A New Decisional Balance Measure of Motivation to Change Among At-Risk College Drinkers

Acknowledgement: Susan E. Collins’s time was supported by National Institute on Alcohol Abuse and Alcoholism (NIAAA) Institutional Training Grant T32AA007455 awarded to Mary E. Larimer at the University of Washington. This study was supported in part by NIAAA Grant AA12518 awarded to Kate B. Carey.

A big thanks goes to Dan J. Neal for his statistical insights and advice, to Carrie Luteran for assembling the data sets, and Sonia Kaur for her help with data entry. Thanks also to Sandra Eck for the helpful ongoing discussions about decisional balance measurement in new participant populations.

Decision making often involves consideration of a set of possible behavioral options and an evaluation of the consequences of each option. These consequences may range from desirable to undesirable effects, also referred to as pros and cons (Janis & Mann, 1977). A decisional balance has thus been operationalized as a representation of the pros and cons of a certain behavior and its potential alternatives.

The explicit consideration of the decisional balance—either as a written or counselor-facilitated exercise—was originally proposed to reduce decision-making errors by making people more cognizant of the decision-making process and the factors contributing to their decisions (Janis, 1968). In this context, decisional balance was designed as a therapeutic exercise to facilitate the complete and realistic assessment of the net value of a potential behavior. As this technique has evolved over time, clinicians have started to employ guided decisional balance exercises as a means of enhancing motivation to change risky health behaviors (Dimeff, Baer, Kivlahan, & Marlatt, 1999; Miller, 1999). Specifically, clients articulate and examine ambivalence about their current behavior to determine whether the weight of the evidence is accumulating toward the need for behavior change (Miller, 1999). In the college drinking literature, a limited number of studies have used a guided, open-ended decisional balance as an intervention tool, but the results of guided interventions have been mixed with regard to alcohol use outcomes (Carey, Carey, Maisto, & Henson, 2006; S. E. Collins & Carey, 2005; LaBrie, Pedersen, Earlywine, & Olsen, 2006).

Alternatively, the data generated in a decisional balance could reflect resolve to enter into a course of action (Janis & Mann, 1977), thus serving as a proxy for motivation to change. This idea was echoed in the work of DiClemente and colleagues (e.g., DiClemente et al., 1991), who have since recognized decisional balance as a marker for the initiation of different stages of change (Pollak, Carbonari, DiClemente, Niemann, & Mullen, 1998). In fact, the stages of change have been operationalized algorithmically as a function of change in the decisional balance (Hall & Rossi, 2008; Prochaska, 1994). Thus, the historical development of the decisional balance construct shows its potential to both enhance and reflect people’s motivational states. In this study, we focused on the role of decisional balance as an assessment tool rather than as an intervention procedure.

The existence of a relationship between decisional balance and motivation to change is evident; however, the nature of this relationship has been described differently in different theoretical contexts. Furthermore, how decisional balance is measured has included the consideration of the pros and cons of a behavior or the pros and cons of an alternative behavior. What exactly decisional balance measures, however, has been less clearly defined. For example, some researchers have asserted that it represents the decision-making process itself (Fischoff & Quadrel, 1991; Janis & Mann, 1977), others have posited that it represents one dimension of motivation to change (Miller, 1999), and yet others have suggested that it is a covariate or perhaps a mechanism involved in transitioning through various stages of behavior change (Prochaska et al., 1994). Considering the ambiguity in the relationship between motivation and decisional balance, more research is needed to establish the validity and clinical utility of decisional balance as a measure of motivation to change.

Reflecting the ambiguity surrounding the decisional balance construct, several decisional balance measures for drinking have been designed over the past 2 decades. The Alcohol Decisional Balance Scale assesses the pros and cons of maintaining one’s current alcohol use using a 42-item Likert scale questionnaire (King & DiClemente, 1993). The Alcohol and Drug Consequences Questionnaire is a 28-item, 6-point Likert scale questionnaire designed to assess the pros and cons of changing alcohol or drug behavior (Cunningham, Gavin, Sobell, Sobell, & Breslin, 1997). Finally, the Decisional Balance for Immoderate Drinking (DBID) was developed for college students (Migneault, Velicer, Prochaska, & Stevenson, 1999). This measure consists of 25 Likert scale questions assessing on a 5-point scale the importance of selected pros and cons of “immoderate” drinking. Different sets of these original 25 items may be summed to form either two 20-item scales (pros vs. cons of immoderate drinking) or three 18-item scales (pros, potential cons, and actual cons).

Despite initially promising psychometric evaluations of these questionnaires, conceptual weaknesses can be identified. First, all of these measures have focused exclusively on the pros and cons of current drinking or the pros and cons of reducing or changing drinking, which precludes the evaluation of the decisional balance as a whole. An incomplete decisional balance has been viewed as problematic in decision-making theory because of the potential for overlooked consequences to create new ambivalence after a decision has been reached (Janis & Mann, 1977). Similarly, an incomplete measure of decisional balance may fail to take into account all aspects of a person’s current motivation to change. One empirical study found that consideration of both the target behavior and an alternative behavior nearly doubled the number of pros and cons spontaneously produced across multiple risky behaviors (Beyth-Marom, Austin, Fischhoff, Palmgren, & Jacobs-Quadrel, 1992). Particularly among adolescents in that study, negative social consequences were reported more often in assessing the cons of not drinking. Thus, a focus on only one half of the decision-making process (i.e., either the pros and cons of current or alternative behavior) might lead to an incomplete and potentially less predictive measure of decisional balance.

Another problem with decisional balance measures to date is that the pros and cons are generated by researchers instead of by participants themselves. This approach may be inadequate in capturing motivation to change authentically and accurately. First, if researchers approach the topic from an academic perspective, they may identify different pros and cons of current drinking versus drinking behavior change than participants (Fischoff & Quadrel, 1991). Also, the language used by researchers to describe the pros and cons can, in and of itself, influences participants’ interpretation of the item and thus the participants’ answers (Beyth-Marom et al., 1992; Fischoff & Quadrel, 1991). On a related point, in providing participants with the “correct” pros and cons of drinking, researchers may be artificially constructing the decision-making process to which participants passively respond. This approach may have the unwanted side effect of making respondents aware of pros and cons they may not have otherwise considered and that may not represent their own unique decision-making process. In contrast, use of an open-ended response format allows participants to express their actual motivational state rather than respond to researchers’ perspectives and values (Fischoff & Quadrel, 1991).

In light of these concerns, it is plausible that an open-ended, participant-generated decisional balance could provide a more accurate measure of motivation to change and better predict drinking outcomes among college drinkers. In an exploratory analysis, Collins and Carey (2005) presented some evidence that, when used as an assessment of motivation, the pattern of responses generated during a decisional balance exercise predicted drinking outcomes with modest success. Specifically, among participants receiving an in-person decisional balance exercise, a greater proportion of pros to cons of changing one’s drinking predicted short-term drinking outcomes.

The current study was designed to extend these preliminary findings and examine an expanded, open-ended, four-field decisional balance worksheet as a measure of motivation to change drinking among at-risk college drinkers. The decisional balance worksheet prompted respondents to report the pros and cons of their current drinking versus reduced drinking. Next, the number of pros and cons reported in each field was converted into a proportion representing the decisional balance toward change or the decisional balance proportion (DBP).

The goal of this study was to test the validity of the DBP as a new measure of motivation to change drinking. We hypothesized that the DBP would evince convergent validity by positively correlating with an alternative, Likert scale measure of readiness to change, and positively correlating with the cons and negatively correlating with the pros of drinking as measured by continuous decisional balance scales. Furthermore, we hypothesized that the DBP would evince discriminant validity by showing nonsignificant correlations with measures assessing dissimilar constructs (i.e., social desirability and demographic variables). Finally, we hypothesized that DBP change scores would evince predictive validity. Initial increases in the decisional balance reflecting movement toward drinking behavior change would predict greater decreases in heavy drinking indices over the follow-up period compared with a DBP reflecting no movement or movement away from change.

Participants

Participants consisted of 143 undergraduate volunteers who had participated in a randomized clinical trial of two types of brief motivational interventions (see Carey et al., 2006). Inclusion criteria for this trial were (a) reporting at least one heavy drinking episode in an average week or at least four heavy drinking episodes in the past month, (b) being 18–25 years of age, (c) being a freshman, sophomore, or junior in college, and (d) consenting to participate. Only students who had participated in the first year of the larger study were included in the current secondary analyses because the format of the original measure was changed after the first year. The new format allowed for electronic data scanning but, as a consequence, limited the number of potential entries. Because the number of entries in each field of the decisional balance was the primary focus of the current study, the data set includes only the participants who had the opportunity to respond to the unrestricted format.

The subsample of 143 students in the current study was predominately female (68%, n = 97), and the average age was 19.20 years (SD = 0.87). The sample consisted of 65% freshmen, 27% sophomores, and 7% juniors. The majority self-identified as White (87%), whereas 2% self-identified as Black/African American, 5% as Asian/Pacific Islander, 3% as Hispanic/Latino/a, and 4% as other or multiracial. Membership in a fraternity or sorority was reported by 24% of the sample, and most participants reported living on campus (89%) or in a fraternity or sorority house (2%), as opposed to off campus (8%).

Measures

A set of demographic questions assessed participants’ age, gender, year in college, ethnicity, on- or off-campus residence, and membership in an on-campus Greek organization. Social desirability was measured using a 13-item short form of the Marlowe Crowne Social Desirability Scale (Reynolds, 1982). The alpha in the current sample was adequate (α = .63).

The decisional balance worksheet was modeled after a scale used to assess the accessibility of alcohol expectancies (see Stacy, Leigh, & Weingardt, 1994). Because it is a free-recall task, it was administered prior to other Likert scale measures of decisional balance. Participants recorded each “advantage” and “disadvantage” of “continuing to drink as you are now” and “drinking less than you do now” on prenumbered lines on the open-ended decisional balance worksheet. The counts of the pros and cons were obtained by summing the prenumbered lines filled in by participants and formed the main explanatory variable for this study, DBP, which may be written as   where subscripts red = reducing drinking and cur = current drinking. DBP scores at 0.5 represent an even balance between pros and cons of reducing drinking and current drinking. Scores between 0.5 and 1.0 indicate a balance tipped toward reducing drinking, and DBP scores between 0.0 and 0.5 indicate a balance tipped toward maintenance of current drinking. In the current study, baseline DBP was used in the convergent and discriminant validity tests, whereas change in the DBP from baseline to 1-month follow-up was used in predictive validity tests.

The 12-item Readiness to Change Questionnaire (RTCQ; Rollnick, Heather, Gold, & Hall, 1992) was scored as a continuous measure of readiness to change (Budd & Rollnick, 1996). The RTCQ was used in convergent and predictive validity analyses in the current study. The alpha reached an adequate level of item consistency (α = .80).

The DBID (Migneault et al., 1999) is a 25-item Likert scale measure developed for college students. Different sets of these original 25 items may be summed to form either two scales (10-item Pros vs. 10-item Cons of immoderate drinking) or three scales (10-item Pros, 3-item Potential Cons, and 5-item Actual Cons). The 18-item, three-factor solution was used in tests of convergent and predictive validity in the current study. In the overall sample, reliability was calculated for Pros (α = .81), Potential Cons (α = .71), and Actual Cons (α = .52).

All drinking assessments used the previous 30 days as a uniform time frame and defined a drink as a 10- to 12-oz can or bottle of 4%–5% beer, a 4-oz glass of 12% table wine, a 12-oz bottle or can of wine cooler, or a 1.25-oz shot of 80-proof liquor either straight or in a mixed drink. A modified version of the Daily Drinking Questionnaire (R. L. Collins, Parks, & Marlatt, 1985) allowed for calculation of drinking frequency and quantity per heaviest drinking week. Using this measure, participants also estimated the frequency of heavy episodic drinking (HED), defined as five or more drinks for men and four or more drinks for women on one occasion (Wechsler et al., 2002). This measure yielded three of the alcohol outcome variables.

The Rutgers Alcohol Problem Index (RAPI; White & Labouvie, 1989) consists of 23 items assessing alcohol-related problems and was specifically developed for use with adolescents and young adults. Participants used a Likert scale to indicate how many times in the past 30 days they experienced each problem listed (i.e., 0 = 0 times, 1 = 1–2 times, 2 = 3–5 times, 3 = 6–10 times, 4 = more than 10 times), and a summary score represented severity of problems. Adequate internal consistency was obtained in the overall sample (α = .82).

Procedure

Participants attending group baseline sessions provided informed consent and completed the measures mentioned above; all received course credit for their participation. Those who reported engaging in HED at least 4 times in the past month were invited via telephone to participate in a randomized clinical trial that included various brief intervention and assessment conditions (for details, see Carey et al., 2006). Following the intervention period, participants were invited to attend 1-, 6-, and 12-month in-person follow-up assessments for which they filled out the same questionnaires and received course credit, $20, and $25, respectively.

Preliminary Data Analysis

The distributions of the explanatory and outcome variables were examined for univariate outliers and deviation from the expected distributions. All drinking outcome measures were positively skewed count variables that approximated the negative binomial distribution. There were no extreme univariate outliers (see Table 1 for all means and standard deviations).

Descriptive Statistics for Explanatory and Response Variables (N = 143)

Convergent Validity

To determine the convergent validity of the DBP, we conducted bivariate Spearman correlations between the baseline DBP and baseline summary scores for the RTCQ, DBID–Pros, DBID–Potential Cons, and DBID–Actual Cons. Although the correlations with the DBP were all in the expected directions, only the correlations between the DBP and RTCQ and DBID–Actual cons were statistically significant (see Table 2 for correlations).

Bivariate Correlations Between the DBP, RTCQ, and DBID Scales

Discriminant Validity

Bivariate Spearman correlations and Mann–Whitney U tests were conducted to test the hypothesized lack of association between the baseline DBP and gender, U(N = 143) = –0.26, p = .80; race/ethnicity, U(N = 143) = –1.32, p = .19; housing situation, U(N = 142) = 0.78, p = .44; social desirability, ρ = –0.09, p = .28; age, ρ = –0.12, p = .14; and Greek membership, U(N = 143) = –1.03, p = .30. As all tests were nonsignificant, discriminant validity of this measure was supported.

Predictive Validity of Change in Decisional Balance

Analysis plan

Population-averaged generalized estimating equation (Zeger & Liang, 1986) models were conducted using STATA 10 (StataCorp, 2007) and tested the change in DBP from baseline to 1-month follow-up as a predictor of heavy drinking outcomes over the 12-month follow-up period. For those unfamiliar with population-averaged generalized estimating equation models, they may be conceptualized as marginal regression models that can be applied to data conforming to different types of distributions (e.g., normal, Poisson, negative binomial, binomial) and can take into account nonindependence resulting from data clustering (e.g., longitudinal data collected on one participant).

The outcome variables were based on a time frame of the previous 30 days and included drinking quantity and frequency during the heaviest drinking week, HED frequency, and RAPI score. Because the distributions of the drinking outcome variables were positively skewed, and the variables were typically overdispersed count/integer responses (e.g., number of drinks consumed during the heaviest drinking week), negative binomial distributions were specified (cf. Neal & Simons, 2007). To enhance interpretability of the regression coefficients, we used the log link for these models. For all variables, repeated measures on one case served as the clustering variable. Because the drinking outcome variables were longitudinal, unevenly spaced, and variably correlated, we used an unspecified correlation structure (Hardin & Hilbe, 2003). The unspecified correlation structure allowed for the correlation in the drinking outcome data at each time point to be taken into account in the overall model estimation.

Three separate models for each of the drinking outcome variables were used to test the relative predictive abilities of DBP, RTCQ, and DBID change scores. The DBP models included five predictors: a linear time variable that compared drinking outcomes at baseline, 1-, 6-, and 12-month follow-ups (coded as 0, 1, 6, and 12, respectively); a quadratic time variable, which took into account the fact that alcohol use over time often follows a curvilinear versus a straight linear path; the DBP change score, which reflected movement in the balance toward or away from change (1-month follow-up minus baseline DBP); and both Linear and Quadratic Time × DBP interactions. Similar but nonnested models involving change on the RTCQ scale and on the three DBID scales were also run and were subsequently compared on goodness of fit using the quasi-likelihood under the independence model information criterion (QICu; Hardin & Hilbe, 2003). Similar to the Akaike’s information criterion tests, statistically superior models have the lowest QICu scores (Hardin & Hilbe, 2003).

Quantity: Heaviest drinking week

The DBP model (QICu = 5124.78) for quantity during the heaviest drinking week was significant, Wald χ²(5, N = 143) = 14.16, p = .01, and statistically superior to both the RTCQ (QICu = 14458.37) and DBID (QICu = 17711.27) models. After controlling for time and baseline DBP, there were significant Linear Time × DPB (IRR = 0.78, SE = 0.07, p = .006) and Quadratic Time × DBP (IRR = 1.02, SE = 0.007, p = .007) interactions. As shown in Figure 1, all participants seemed to decrease their heavy drinking quantity between baseline and 1-month follow-up, possibly as a result of assessment reactivity. However, increases in DBP scores over the initial 1-month period predicted subsequent decreases in drinking quantity at the 6-month follow-up, an effect that decayed by the 12-month follow-up. Decreasing DBP scores tended to predict an increase in drinking quantity, which was followed by a reduction at the 12-month follow-up. On the other hand, relatively stable DBPs were associated with little change in drinking quantity.

Figure 1. Graph of mean quantity per heaviest drinking week by time point and level of change in decisional balance proportion (DBP). The DBP change scores represent change in the DBP between baseline and the 1-month follow-up. For clarity of presentation, groups were formed to represent different levels of change in the DBP. The stable DBP group in the figure is centered on the mean DBP change score (M = –0.004, SD = 0.20) in this sample and includes difference scores ranging from –0.20 to 0.20. These scores correspond to 1 SD below and 1 SD above the mean, respectively. The decreasing DBP group represents participants whose DBP change scores were at least 1 SD below the mean (DBP < –0.20), and the increasing DBP group represents participants whose DBP change scores were at least 1 SD above the mean (DBP > 0.20).

Frequency: Heaviest drinking week

The DBP model (QICu = 343.23) for frequency during the heaviest drinking week was statistically superior to both the RTCQ (QICu = 1100.58) and DBID (QICu = 1280.73) models, and was significant, Wald χ²(5, N = 143) = 15.36, p = .009. After controlling for time and baseline DBP, there were significant Linear Time × DPB (IRR = 0.84, SE = 0.05, p = .003) and Quadratic Time × DBP (IRR = 1.01, SE = 0.005, p = .005) interactions. After all groups initially decreased on heavy drinking frequency, increasing DBP over the initial 1-month period predicted greater decreases in drinking frequency during the heaviest drinking week—until the 12-month follow-up, when this decreasing effect decayed (see Figure 2). Decreasing DBP scores tended to predict increases in drinking frequency up to the 6-month time point, followed by a reduction at the 12-month follow-up. On the other hand, relatively stable DBPs were associated with little change in drinking frequency.

Figure 2. Graph of mean frequency per heaviest drinking week by time point and level of change in decisional balance proportion (DBP).

HED

The best model according to the QICu was the DPB model (QICu = 1481.03) compared with the RTCQ (QICu = 4069.71) and DBID (QICu = 4891.80) models; however, none of the omnibus model tests for HED outcomes were significant (all ps > .18).

RAPI

Compared with the RTCQ (QICu = 5375.68) and DBID (QICu = 6493.62) models, the DBP (QICu = 1975.94) model provided the best fit for self-reported alcohol-related problems, Wald χ²(5, N = 143) = 13.12, p = .02. After controlling for time and baseline DBP, there was a significant Linear Time × DPB interaction (IRR = 0.85, SE= 0.07, p = .046). As shown in Figure 3, all participants reported initial decreases in RAPI. Increases in DBP over the initial 1-month period, however, predicted a stable linear decrease in alcohol-related problems over the follow-up period. On the other hand, decreasing DBP scores tended to predict increases in alcohol-related problems up to the 6-month time point, followed by a downward trend at the 12-month follow-up. On the other hand, relatively stable DBPs are associated with relatively stable experience of alcohol-related problems.

Figure 3. Graph of mean alcohol-related problems (Rutgers Alcohol Problem Index [RAPI] score) by time point and level of change in decisional balance proportion (DBP).

This study provided an examination of a DBP as a measure of motivation to change among at-risk college drinkers. The DBP was generated from responses to an open-ended decisional balance worksheet assessing pros and cons of current drinking versus reducing drinking; it was constructed to reflect the extent to which the decisional balance was tipped toward change.

Convergent validity of the DBP was partially supported in this study. As predicted, initial DBP positively and significantly correlated with readiness to change as measured by the RTCQ. This finding provided convergent validity for the DBP as a measure of motivation to change. Furthermore, the weighted importance of current negative outcomes (DBID–Actual Cons) in participants’ decisions to drink was significantly, albeit weakly, associated with DBP scores. The somewhat weak effect may indicate that the DBID and the DBP measure overlapping yet distinct constructs, or it may reflect the relatively low reliability of the DBID–Actual Cons scale. The latter point is a psychometric issue that may have limited the power to optimally assess convergent validity with this scale.

The DBP was not significantly correlated with the pros scale of the DBID, which taps into the importance of positive aspects of “immoderate” drinking (e.g., “I feel happier when I drink”). However, this nonsignificant correlation is understandable: The DBP was constructed to reflect the tilt of the decisional balance toward change, not the status quo. In fact, the DBP represents the weight of the cons of current drinking plus the pros of changing relative to all fields in the balance. Because this proportion does not explicitly highlight the pros of current drinking, they may be “passively” outweighed.

The fact that the DBP was not correlated with the DBID–Potential Cons scale reflects the mixed findings regarding convergent validity of the DBP. However, this lack of association may also be interpreted in the context of the cognitive and memory literature, which asserts that individual drinking experience influences the accessibility of certain thoughts about alcohol use. Specifically, frequently encountered outcomes, such as those represented by the DBID–Actual Cons scale (e.g., “Drinking makes me feel out of control”), may be more accessible than hypothetical, potential outcomes, such as those measured by the DBID–Potential Cons scale (e.g., “Drinking could kill me”; Stacy et al., 1994). On the other hand, the DBP, which reflects an open-ended assessment of the decisional balance, may be even more accurate in assessing the most personally salient pros and cons of a behavior and may be more reflective of an individual’s current motivational state than the DBID scales.

Evidence for the discriminant validity of the DBP was obtained. Our findings confirmed that the DBP did not significantly correlate with demographic measures (i.e., gender, race/ethnicity, year in college, Greek membership, age) or social desirability. Perusal of the recent literature reveals no studies involving college students that have demonstrated associations between motivation and basic demographic variables or social desirability. Thus, consistent with empirical precedent and theory, the DBP appears to be an independent construct that may be used with a range of college students.

Findings in this study revealed that changes in the DBP predicted heavy drinking outcomes among at-risk college drinkers. In fact, the DBP models fit the data better than did models including RTCQ and DBID change scores as predictors of alcohol use over time. Furthermore, the DBP models were the only ones that yielded consistently significant predictors of drinking outcomes. Across three of the four drinking outcome variables, movement toward change (i.e., an increase in proportion of pros of reduced drinking and cons of current drinking to total item count from baseline to 1-month follow-up) predicted reductions in drinking over the initial follow-up period.

However, this finding was tempered by a curvilinear effect on two of the four drinking outcome variables, drinking quantity and frequency, which complicated the initial linear findings. After participants with increasing motivation to change initially decreased their heavy drinking quantity and frequency (and vice versa), there was an apparent decay in this effect at the 12-month follow-up. This regression to the mean, however, does not necessarily indicate that the DBP does not predict drinking as hypothesized. It is highly possible that the relatively short-term changes in the DBP that occurred initially from baseline to 1-month posttest may be most helpful in predicting more proximal changes in drinking. Thus, the more distal drinking measured 1 year after the initial motivation ratings may not be as reliably predicted by the DBP. This explanation fits with the literature on motivation, which suggests that motivation to change drinking is a fluid state rather than a stable trait (Miller, 1999). It also corresponds to the developmental literature that documents temporal variation in college student drinking patterns over weeks and years (Del Boca, Darkes, Greenbaum, & Goldman, 2004; Schulenberg, O’Malley, Bachman, Wadsworth, & Johnston, 1996). Further studies are needed to establish the temporal robustness of changes in motivation as measured by the DBP. Perhaps models assessing the time-varying and parallel change in DBP and drinking would be a helpful next step.

Conceptually, the DBP corresponds to decisional balance and motivation to change theory better than the other measures (i.e., RTCQ and DBID) tested as predictors in this study. Unlike previous studies involving decisional balance measures (e.g., Cunningham et al., 1997; King & DiClemente, 1993; Migneault et al., 1999; Velicer, DiClemente, Prochaska, & Brandenburg, 1985), the DBP integrates all four fields of the decisional balance: the pros and cons of both current drinking and drinking reduction. Because both theory and empirical findings have indicated that the consideration of the pros and cons of both current behavior as well as behavior change are key to accurate assessment of a person’s current motivation to change (Beyth-Marom et al., 1992; Janis & Mann, 1977), the use of the decisional balance worksheet may represent a step forward in decisional balance measurement. Furthermore, the fact that the input for the DBP is participant- instead of researcher-generated may make this measure a more accurate and personally relevant representation of one’s motivation to change than the RTCQ and DBID (Fischoff & Quadrel, 1991). Finally, the open-ended format of the decisional balance worksheet lends itself to potential qualitative as well as quantitative representations of motivation to change.

Limitations

This study comprised a nonrandom sample of at-risk college drinkers who had participated in a larger intervention trial. Considering the potential confounding effects of the nonrandom selection and exposure to brief interventions, it is necessary to replicate these results on a larger, randomly selected, nontreatment sample. Furthermore, the relatively homogeneous racial and ethnic composition of the current sample raises questions as to the external validity of the current findings. This sample consisted of predominantly White, non-Hispanic students; thus, further study of the DBP and its ability to predict drinking outcomes in more diverse samples is necessary to ensure its generalizability to other populations. We also recognize that the DBP focuses on numbers of items generated rather than their content. Although potential information may be gained by considering item content as well, the DBP has the advantage of rapid and reliable scoring. Despite these limitations, the current results provide additional support for and expansion on a quantification of a drinking decisional balance originally introduced by Collins and Carey (2005).

Conclusions

This study has provided evidence for the convergent, discriminant, and predictive validity of a new decisional balance measure of motivation to change drinking behavior among at-risk college drinkers. This study adds to the literature because previous conceptualizations of the decisional balance measure were researcher- instead of participant-generated and were limited in their scope (i.e., assessed either pros and cons of drinking or of changing behavior but not both). Furthermore, this measure appears to predict longitudinal drinking better than established, Likert scale measures of readiness to change and decisional balance. Larger scale studies should be conducted to provide additional support for the psychometric integrity, clinical utility, and generalizability of this decisional balance measure.

Beyth-Marom, R., Austin, L., Fischhoff, B., Palmgren, C., & Jacobs-Quadrel, M. (1992). Perceived consequences of risky behaviors: Adults and adolescents. Developmental Psychology, 29, 549–563.

Budd, R. J., & Rollnick, S. (1996). The structure of the Readiness to Change Questionnaire: A test of Prochaska & DiClemente’s transtheoretical model. British Journal of Health Psychology, 1, 365–376.

Carey, K. B., Carey, M. P., Maisto, S. A., & Henson, J. M. (2006). Brief motivational interventions for heavy college drinkers: A randomized controlled trial. Journal of Consulting and Clinical Psychology, 74, 943–954.

Collins, R. L., Parks, G. A., & Marlatt, G. A. (1985). Social determinants of alcohol consumption: The effects of social interaction and model status on the self administration of alcohol. Journal of Consulting and Clinical Psychology, 53, 189–200.

Collins, S. E., & Carey, K. B. (2005). Lack of effect for decisional balance as a brief motivational intervention for at-risk college drinkers. Addictive Behaviors, 30, 1425–1430.

Cunningham, J. A., Gavin, D., Sobell, L. C., Sobell, M. B., & Breslin, F. C. (1997). Assessing motivation for change: Preliminary development and evaluation of a scale measuring the costs and benefits of changing alcohol or drug use. Psychology of Addictive Behaviors, 11, 107–114.

Del Boca, F. K., Darkes, J., Greenbaum, P. E., & Goldman, M. S. (2004). Up close and personal: Temporal variability in the drinking of individual college students during their first year. Journal of Consulting and Clinical Psychology, 72, 155–164.

DiClemente, C. C., Prochaska, J. O., Fairhurst, S. K., Velicer, W. F., Velasquez, M. M., & Rossi, J. S. (1991). The process of smoking cessation: An analysis of precontemplation, contemplation, and preparation stages of change. Journal of Consulting and Clinical Psychology, 59, 295–304.

Dimeff, L. A., Baer, J. S., Kivlahan, D. R., & Marlatt, G. A. (1999). Brief alcohol screening and intervention for college students (BASICS): A harm reduction approach. New York: Guilford Press.

Fischoff, B., & Quadrel, M. J. (1991). Adolescent alcohol decisions. Alcohol Health and Research World, 15, 43–51.

Hall, K. L., & Rossi, J. S. (2008). Meta-analytic examination of the strong and weak principles across 48 health behaviors. Preventive Medicine, 46, 266–274.

Hardin, J. W., & Hilbe, J. M. (2003). Generalized estimating equations. Boca Raton, FL: Chapman & Hall/CRC.

Janis, I. L. (1968). Pilot studies on new procedures for improving the quality of decision making. New Haven, CT: Yale Studies in Attitudes and Decisions.

Janis, I. L., & Mann, L. (1977). Decision-making: A psychological analysis of conflict, choice, and commitment. New York: The Free Press.

King, T. K., & DiClemente, C. C. (1993, November). A decisional balance measure for assessing and predicting drinking behavior. Paper presented at the Annual Conference of the Association for the Advancement of Behavior Therapy, Atlanta, GA.

LaBrie, J. W., Pedersen, E. R., Earlywine, M., & Olsen, H. (2006). Reducing heavy drinking in college males with the decisional balance: Analyzing an element of motivational interviewing. Addictive Behaviors, 31, 254–263.

Migneault, J. P., Velicer, W. F., Prochaska, J. O., & Stevenson, J. F. (1999). Decisional balance for immoderate drinking in college students. Substance Use & Misuse, 34, 1325–1346.

Miller, W. R. (1999). Enhancing motivation for change in substance abuse treatment (Vol. 35). Rockville, MD: U.S. Department of Health and Human Services.

Neal, D. J., & Simons, J. S. (2007). Inference in regression models of heavily skewed alcohol use data: A comparison of ordinary least squares, generalized linear models, and bootstrap resampling. Psychology of Addictive Behavior, 21, 441–452.

Pollak, K. I., Carbonari, J. P., DiClemente, C. C., Niemann, Y. F., & Mullen, P. D. (1998). Causal relationships of processes of change and decisional balance: Stage-specific models for smoking. Addictive Behaviors, 23, 437–448.

Prochaska, J. O. (1994). Strong and weak principles for progressing from precontemplation to action on the basis of twelve problem behaviors. Health Psychology, 13, 47–51.

Prochaska, J. O., Velicer, W. F., Rossi, J. S., Goldstein, M. G., Marcus, B. H., Rakowski, W., et al. (1994). Stages of change and decisional balance for 12 problem behaviors. Health Psychology, 13, 39–46.

Reynolds, W. M. (1982). Development of reliable and valid short forms of the Marlowe–Crowne Scale of Social Desirability. Journal of Clinical Psychology, 38, 119–125.

Rollnick, S., Heather, N., Gold, R., & Hall, W. (1992). Development of a short “Readiness to Change Questionnaire” for use in brief, opportunistic interventions among excessive drinkers. British Journal of Addictions, 87, 743–754.

Schulenberg, J., O’Malley, P. M., Bachman, J. G., Wadsworth, K. N., & Johnston, L. D. (1996). Getting drunk and growing up: Trajectories of frequent binge drinking during the transition to young adulthood. Journal of Studies on Alcohol, 57, 289–304.

Stacy, A., Leigh, B. C., & Weingardt, K. R. (1994). Memory accessibility and association of alcohol use and its positive outcomes. Experimental and Clinical Psychopharmacology, 2, 269–282.

StataCorp. Stata statistical software (Release 10) [computer software]. College Station, TX: StataCorp.

Velicer, W. F., DiClemente, C. C., Prochaska, J. O., & Brandenburg, N. (1985). Decisional balance measure for assessing and predicting smoking status. Journal of Personality and Social Psychology, 48, 1279–1289.

Wechsler, H., Lee, J. E., Kuo, M., Seibring, M., Nelson, T. F., & Lee, H. (2002). Trends in college binge drinking during a period of increased prevention efforts. Journal of American College Health, 50, 203–217.

White, H. R., & Labouvie, E. W. (1989). Towards the assessment of adolescent problem drinking. Journal of Studies on Alcohol, 50, 30–37.

Zeger, S. L., & Liang, K.-Y. (1986). Longitudinal data analysis for discrete and continuous outcomes. Biometrics, 42, 121–130.

Submitted: October 1, 2008 Revised: January 1, 2009 Accepted: February 20, 2009

A Tutorial on Count Regression and Zero-Altered Count Models for Longitudinal Substance Use Data

Acknowledgement: We thank Eun-Young Mun, Zac Imel, Isaac Rhew, Jennifer Kirk, and an anonymous reviewer for helpful feedback that improved the manuscript in numerous ways. The example data included in the article were collected via the National Institute on Alcohol Abuse and Alcoholism (NIAAA) Grants AA016099 and AA014576 (Clayton Neighbors, PI), and David C. Atkins' time was supported in part by NIAAA Grant AA019511 (Eun-Young Mun, PI).

What is the impact of personalized normative feedback on drinking related problems in college students over time? How does weekend versus weekday drinking vary by gender and fraternity/sorority status when assessed on a daily basis? Many questions about alcohol and substance use focus on change across time, and the methods used to analyze these questions need to account for the longitudinal nature of the data. Generalized linear mixed models (GLMMs; Gelman & Hill, 2007; Hedeker & Gibbons, 2006; also called hierarchical [or multilevel] generalized linear modeling, Raudenbush & Bryk, 2002; Snijders & Bosker, 1999) are increasingly common analytic approaches for longitudinal data, given their flexible handling of unbalanced repeated measures (i.e., individual participants may have unique numbers and timings of assessments) and the widespread availability of software for estimating such models. Moreover, GLMMs are appropriate for continuous as well as discrete outcomes.

However, the distributions of alcohol and substance abuse outcomes have characteristic shapes: They are often positively skewed and bounded by zero. Moreover, there can be a large stack of data points at zero, indicating individuals and/or occasions without drinking, use, or related problems. These distributions reflect that alcohol and substance use outcomes are often count data, representing a total number of something, be it drinks, days using, or number of problems. Except in special circumstances (e.g., specially selected samples with high drinking or drug use), statistical models that assume normally distributed residuals will provide poor fit to such data and will lead to incorrect confidence intervals and p values. Instead, count regression approaches such as Poisson or negative binomial regression or zero-altered count models (e.g., zero-inflated or hurdle models) are much more appropriate for these types of data (Atkins & Gallop, 2007; Coxe, West, & Aiken, 2009; Hilbe, 2011; Neal & Simons, 2007; Simons, Neal, & Gaher, 2006).

In the past, addictions researchers have often ignored (or not been aware of) violations of distributional assumptions or have attempted to deal with them in nonoptimal ways. Count regression models are beginning to be applied to addictions data (e.g., Gaher & Simons, 2007; Lewis et al., 2010), but accessible resources on how to apply these models to longitudinal data are scarce. The present article provides a tutorial in analytic methods for count data from longitudinal studies, focusing on extensions to GLMMs for count outcomes. We use two examples from our research to illustrate the need for, and application of, longitudinal count models. Data and computer code to run the analyses in R, Mplus, SAS, Stata, and SPSS are available on a supplementary website (http://depts.washington.edu/cshrb/newweb/statstutorials.html), though note that not all software can run all models that are covered here at present time. The outline of the article is as follows: Introduction to example data and research questions, brief overview of count regression models, GLMMs for count regression models, analyses and interpretation of example data, and discussion of software and practical issues in using these methods. In addition, there is a technical appendix containing important, but more advanced, material (see online supplemental material; http://dx.doi.org.offcampus.lib.washington.edu/10.1037/a0029508.supp). We assume that readers have a basic familiarity with linear mixed models (i.e., hierarchical linear or multilevel models assuming normally distributed errors) and count regression models, though both are introduced briefly here and introductory resources are highlighted throughout.

The first example dataset is drawn from an intervention study aimed at reducing problematic drinking in college students (Neighbors et al., 2010). The current article focuses on gender differences across 2 years in alcohol-related problems, as measured by the Rutgers Alcohol Problem Index (RAPI; White & Labouvie, 1989). The dataset includes 3,616 repeated measures across five time points from 818 individuals. The second dataset involves intensive, daily assessment of drinking. The data come from a larger intervention study of event-specific prevention (i.e., drinking related interventions for 21st birthdays and spring break), but the current data are observational. Specifically, these data record the number of drinks for each day over approximately the last 30 days for 980 individuals (23,992 total person-days), as measured by the timeline follow-back interview (TLFB). These data came from a survey study of 21st birthday drinking and consequently include some extreme drinking events relative to a random sample of student's drinking (Neighbors et al., 2011). Analyses focus on drinking differences by gender, Greek status (i.e., fraternity/sorority membership), and weekend (Thurs–Sat) versus weekday (Sun–Wed). (Note that Thursday was included as part of the “weekend” given that drinking on Thursday was more similar to Friday and Saturday drinking than other days of the week.)

Count variables are often positively skewed and often include many observations at zero. The top row of Figure 1 displays (unconditional) distributions of drinking (TLFB) and alcohol problems (RAPI), which are strongly skewed with a mode of zero. The bottom row of Figure 1 shows histograms of residuals from regressing the RAPI on gender and time assuming normally distributed residuals (i.e., ordinary least squares [OLS] regression): on the left without any transformation and on the right with a log transformation of the RAPI. These plots show that skewed count regression outcomes will rarely meet the distributional assumptions of OLS regression or linear mixed models. Moreover, count outcomes will also typically violate the equal variances (i.e., homoskedasticity) assumption of linear models as count outcomes have a direct relationship between their mean and variance, where higher levels of the outcome have greater variance. Although transforming the outcome is a commonly suggested strategy for skewed data, a stack of zeroes will not be smoothed out by a transformation. Moreover, focusing on the TLFB data, the distribution shown in Figure 1 is suggestive of two different types of associations or research questions: (a) what is related to no drinking versus any drinking (i.e., zero vs. nonzero), and (b) what is related to the amount of drinking when there is drinking? Although the zero/nonzero aspects of the RAPI data are not quite as notable, a similar set of questions could be asked of the RAPI data. As will be described later, zero-inflated and hurdle models have submodels that focus on these two questions. Prior to introducing count regression models, we consider the qualities of the example data that are related to the questions just posed.

Figure 1. Plots of frequency counts of daily drinks from timeline follow-back (TLFB; upper left) and Rutgers Alcohol Problems Index (RAPI; upper right). Residuals from fitting an ordinary least squares regression to the RAPI or log-transformed RAPI are in lower left and lower right, respectively.

Consider how the proportion of individuals drinking versus not drinking and number of drinks on drinking days are related to covariates in the TLFB data. Figure 2 presents means and 95% confidence intervals (CIs) for number of drinks on drinking days (top half of graph) and proportion of days drinking (bottom half of graph) by weekday versus weekend, Greek status, and gender.

Figure 2. Means and 95% confidence intervals for drinking on drinking days (top row) and proportion of individuals drinking (bottom row). Means are stratified by weekend versus weekday, male versus female, and fraternity/sorority member or not.

There are strong differences in the proportion of people drinking for weekend versus weekday, with less prominent differences for number of drinks when drinking. Moreover, from this descriptive view, the number of drinks is highly related to fraternity/sorority status, whereas proportion of people drinking does not appear as strongly related to Greek status.

Figure 3 shows a similar set of plots for the RAPI by gender across time. As with the TLFB drinking data, patterns of association appear different across the two aspects of the outcome. Over the five assessments (and following intervention for most participants), the proportion of individuals with any alcohol problems is dropping, with a greater proportion of men consistently reporting more alcohol problems (right panel). When examining the number of alcohol problems (given some problems), we see evidence for divergence between the sexes, with women showing slight decreases in number of alcohol problems, whereas men appear to show slight increases, though with notable variability as seen in the confidence intervals (left panel). Thus, these graphs serve to highlight that different associations can occur with outcomes with notable zeroes (i.e., whether there is any drinking [zero vs. not zero] and amount of drinking when any drinking).

Figure 3. Means and 95% confidence intervals for number of problems when there are any problems (left) and proportion of individuals reporting any drinking-related problems (right). Means are stratified by assessment period and gender.

Before moving on to discuss GLMMs for count outcomes, we briefly introduce count regression models (more thorough introductions can be found in articles by Atkins & Gallop, 2007, and Coxe et al., 2009, and the book by Hilbe, 2011). The basic count regression model is Poisson regression, which is one of the generalized linear models (McCullagh & Nelder, 1989). There are two critical differences between OLS regression and Poisson regression. First, the outcome (conditional on covariates) is assumed to be distributed as a Poisson random variable as opposed to a Normal random variable. The top row of Figure 4 depicts three different Poisson distributions, with varying means (denoted by the Greek letter mu). This figure underscores that the Poisson distribution is a discrete distribution for non-negative integers, the exact qualities of count variables (i.e., a count variable cannot be negative or fractional). Second, in Poisson regression the linear predictor of the regression model (i.e., the right-hand side of the regression equation) is connected to the outcome via a natural logarithm link function. Although this is not identical to transforming the outcome, it does mean that the regression coefficients from a Poisson model are on a log scale. Similar to logistic regression, raw coefficients are typically raised to the base e (i.e., exponentiated, the antilog function) and interpreted as rate ratios. Like odds ratios (ORs), rate ratios are inversely proportional around one (i.e., a rate ratio of 3 is equal in strength but opposite in direction to a rate ratio of 1/3). Rate ratio interpretation will be described in greater detail later, in the applications section.

Figure 4. Plots of frequency counts of data simulated from Poisson distributions with three different means (top row). The bottom row contains frequency counts of data simulated from negative binomial distributions with the same mean but varying dispersion.

The Poisson distribution (and regression) has an Achilles' heel of sorts in that it has the property that the mean equals the variance. In real data the variance often far exceeds the mean, and we would say that the data are overdispersed relative to the Poisson distribution. Using descriptive statistics, dispersion is typically defined by the ratio of the variance to the mean. Thus, a Poisson distribution assumes a dispersion parameter of 1 (also called equidispersion). A dispersion parameter of 3 would indicate a variance value of three times the mean, which would (descriptively) indicate overdispersion. Both of our example datasets show evidence of overdispersion simply based on descriptive statistics (TLFB: M = 1.2, Var = 9.3; RAPI: M = 6.3, Var = 82.9), though overdispersion can also be influenced by the longitudinal nature of these data and may be accounted for by covariates included in the model. Note that this description is meant to convey the intuitive ideas underlying overdispersion, but formal tests can (and should) be used in regression modeling of count data (see, e.g., Hilbe, 2011).

The negative binomial model extends the Poisson model by allowing the mean and variance to be different. The lower row of Figure 4 presents three different negative binomial distributions. Note that each of these distributions has the same mean, but the dispersion varies, highlighting the primary difference between the Poisson and negative binomial distributions. The Poisson regression model is a special case of the negative binomial model, and when the mean equals the variance, the two will yield identical results. However, when the variance exceeds the mean, the negative binomial is more appropriate, and its standard errors will be reliably larger than those from the Poisson, reflecting the additional variance in the outcome. In practice, the Poisson model is rarely a good fitting model for exactly this reason, and selecting a Poisson model when the data are overdispersed will yield overly liberal statistical tests (i.e., p values will appear significant when they are not in reality, using a more appropriate model).

As seen in Figure 4, the negative binomial regression model can fit highly skewed data, including data with a relatively large number of zeroes. However, when there is a clear stack of zeroes in the data and especially when the nonzero distribution is not a smooth extension from the zeroes, alternative models may be appropriate. Two closely related models explicitly handle count data with zeroes above and beyond what would be predicted by a negative binomial model: zero-inflated models and hurdle models (see, e.g., Hilbe, 2011; Zeileis, Kleiber, & Jackman, 2008). Both models include a logistic regression for the zeroes in the data and a count regression (either Poisson or negative binomial) for the counts. However, zero-inflated and hurdle models take different approaches to dividing the data around zero. The predicted zeroes in a zero-inflated model come from both the count distribution and an extra mass of zeroes and are a type of mixture model in which the distribution at zero arises from two sources (i.e., count and logistic submodels). Thus, the logistic regression model in a zero-inflated model is for “excess zeroes,” over and above what would be predicted by the count distribution. Hurdle models, on the other hand, model all the zeroes in the logistic regression, and nonzero counts are modeled by a truncated count regression (i.e., truncated because it does not include zero). Hurdle models are related to the more general class of models, often called “two-part models,” in which a logistic model for zero versus nonzero is combined with a model for nonzero values. In the current article we consider only Poisson and negative binomial for nonzero models, but this does not have to be the case (see, e.g., health care costs as in Buntin & Zaslavsky, 2004). In many instances zero-inflated models and hurdle models will yield similar results; however, hurdle models are more straightforward to interpret as all zeroes are handled in one portion of the model, and computationally, hurdle models are somewhat easier to fit as the two parts of the model can be fit independently of one another.

All of the count regression models introduced thus far assume independent observations. This assumption is violated in longitudinal data because repeated observations on the same individual will be correlated. To accommodate this nonindependence of observations, mixed model extensions to count regressions can be used.

The top half of Figure 5 displays change across time in the RAPI for eight randomly selected individuals. As seen in that plot individuals started the study with a wide range of alcohol problems, and some individuals made notable changes in alcohol problems, whereas others did not. To illustrate the variability in intercepts (i.e., initial RAPI) and slopes (i.e., change over time in RAPI), Poisson regressions were fit to each individual's data (i.e., RAPI was regressed on time for each individual separately, using a Poisson regression; see Singer & Willett, 2003, for general discussion of this approach). The distributions of intercepts and slopes are plotted in the bottom of Figure 5. The means of these distributions are descriptive estimates of average intercepts (M = 8.2) and slopes (M = 0.99), and the distributions themselves highlight the variability across individuals in intercepts and slopes.

Figure 5. RAPI scores plotted over time for eight randomly selected individuals (top row). Distribution of intercepts (bottom left) and slopes (bottom right) from fitting separate Poisson regressions to each individual's data.

GLMMs for longitudinal data extend this logic of individual growth curves by including subject-specific variability in one or more components of the model (e.g., intercepts, slopes) via additional variance terms that describe a distribution of regression coefficients across individuals within the study. GLMMs have been referred to as random coefficient models for precisely this reason. Using the systems of equations or hierarchical format with explicit Level 1 and Level 2 models, a Poisson GLMM for the RAPI data might be       which is identical to the composite form of the model by substituting the Level 2 equations into their Level 1 counterparts:   where t indexes time, i indexes individuals, and the linear predictor (i.e., right-hand side of equation) is connected to the mean of the outcome via a natural logarithm link function. Male is a dummy-variable for gender (female = 0, male = 1), and Time measures time in months since the start of the study. The two variance terms (u0i, u1i) describe the deviations of each participant's intercept and slope around the overall trajectory for the sample, defined by the fixed effects (i.e., b0+ b1Male + b2Time). As is common with error terms, these variances are assumed (multivariate) normally distributed. Finally, the model assumes that the conditional distribution of the outcome given fixed and random effects is Poisson distributed. It is important to note that both the fixed and random effects are connected to the outcome via the link function, which has implications for interpretation of the model (see discussion of marginal vs. conditional effects, below).

The results for the model in Equation 1 are presented in Table 1. Focusing first on the fixed effects, the subject-specific rate ratios (RRs) are simply the exponentiated coefficients (e.g., for the intercept, e^1.45 = 4.28). The intercept RR provides the estimated alcohol problems at baseline for women (i.e., when all covariates are equal to zero, as in all regression models), conditional on the random effects. The RR of 1.29 for men indicates that their alcohol problems are 29% higher than women on average. Generally, the distance above or below 1 is interpreted as the percentage increase or decrease in the outcome for a one-unit increase in the predictor. Similarly, the 0.97 RR for time implies a 3% reduction in RAPI for each 1-month increase. Let's consider the interpretation of RRs a bit more closely. The predicted regression lines based on the fixed effects are shown in Figure 6, and the specific values are presented in the figure as well. If this were a linear mixed model or OLS regression, the predicted regression lines would be perfectly straight and parallel (i.e., because there is no interaction between Male and Time), but it is clear from the figure that the predicted regression lines for men and women from the Poisson GLMM are curved and are getting closer to one another over time. Poisson models with their log link functions are sometimes called multiplicative models, whereas OLS regression (and linear mixed models) are considered additive. For the latter, the regression coefficients tell us how much to “add” to our prediction for each one-unit increase in a covariate.

GLMM Poisson Results for RAPI Data

Figure 6. Predicted (unit-specific) RAPI scores from Poisson GLMM for men and women over time, with specific values noted in text.

The interpretation of multiplicative models is more complicated. First, we exponentiate the fixed effects in Equation 1b:   where exp is the exponentiating function, raising to the base e. We then reexpress Equation 1c using a property of exponetiated sums:  

Equation 1d shows why Poisson models are considered multiplicative: The exponentiated coefficients—the RRs—provide a multiplying factor for each unit change in the covariate. Thus, the RR of 1.29 for men indicates that men's predicted alcohol problems are 1.29 times women's predicted alcohol problems at each level of time, which is identical to saying that it is 29% higher (e.g., at time = 0, 4.28 × 1.29 = 5.52, and at time = 24 months, 2.03 × 1.29 = 2.62). For every 6 months, the model predicts alcohol problems are falling by e^−0.03∗6 = 0.83 (i.e., an RR for each 6-month change in time). We can similarly confirm that each 6-month change in predicted alcohol problems is 17% less (i.e., multiply by 0.83) than the preceding predicted value. Thus, for OLS regression and linear mixed models, coefficients describe the amount added to predictions with each one-unit covariate change, whereas Poisson and Poisson GLMM coefficients describe the amount multiplied.

One critical feature of these coefficients that has not been widely noted in the applied literature is that these coefficients (and their interpretation) are conditional on specific values of the random-effects distribution. In linear mixed models, fixed-effect coefficients can be interpreted as “averaging over” the random effects (sometimes called marginal coefficients), but this is not true in GLMM with nonidentity link functions (see, e.g., Heagerty & Zeger, 2000 and Raudenbush & Bryk, 2002, Chapter 10). This issue of conditional versus marginal coefficients (also called unit-specific vs. population-average) is discussed in the online supplemental material (see http://dx.doi.org.offcampus.lib.washington.edu/10.1037/a0029508.supp), along with additional resources.

There are several considerations in judging how well the model fits the data. First, similar to all regression models, we should consider whether we have the correct predictors included in the model and whether any terms have nonlinear associations with the outcome. These decisions should be informed by past research, theory, and thorough descriptive statistics and graphs. In fact, it is hard to overemphasize the importance of thorough descriptive analyses to help guide decisions in building complicated models, such as those considered here. In addition, similar to many regression models, GLMMs assume that variance terms are normally distributed and homoskedastic, and plots of these assumptions can be found in the accompanying online materials (see eFigure 1 in online supplementary material at http://dx.doi.org.offcampus.lib.washington.edu/10.1037/a0029508.supp). Second, similar to linear mixed models we should consider which terms in our model should have corresponding random effects; with GLMMs decisions about which random effects to include are usually determined by testing nested models via deviance tests (also called likelihood ratio chi-square tests; Molenberghs & Verbeke, 2005). Based on descriptive statistics and graphs as well as deviance tests for random effects, the initial model for the RAPI was extended to include an interaction between gender and time and an additional per-observation random effect (discussed below). These results are shown under Model 2 in Table 1.

Most (but not all) GLMMs are fit via a procedure called maximum likelihood, and when the iterative fitting procedure stops, a deviance statistic is reported (technically, −2 × log-likelihood, and sometimes reported as −2LL). The difference between deviances from two nested models is distributed as a chi-square random variate and hence can be tested using a chi-square distribution with degrees of freedom equal to the difference in parameters between the models (see Singer & Willett, 2003, for further details). Deviance tests can also be used with count regression GLMMs, with one major caveat. Fitting GLMMs with non-normal outcomes (e.g., binary or count outcomes) is considerably more challenging, and there are a variety of optimization strategies and algorithms. Only certain algorithms have been shown to be accurate enough to allow deviance tests of nested models. A brief overview of algorithms and approaches is presented in the online supplemental material (see http://dx.doi.org.offcampus.lib.washington.edu/10.1037/a0029508.supp).

There are two final, critical aspects of model evaluation for GLMMs for count outcomes: overdispersion and zeroes. As noted earlier, the Poisson GLMM assumes that, conditional on fixed and random effects, the distribution of the outcome is distributed as a Poisson variable, with its property of equidispersion (i.e., mean equal to the variance). The most straightforward way to examine whether data are overdispersed relative to the Poisson is to fit a model that allows for overdispersed data. Logically, we might consider the negative binomial. For technical details that go beyond our focus here, the negative binomial model is challenging (though not impossible) to extend to random effects (see, e.g., Hilbe, 2011, pp. 488–501), though some software packages are beginning to incorporate the negative binomial with random effects (e.g., Mplus, R, SAS). With GLMMs we can extend the Poisson model to include a per-observation error term, which captures overdispersion. This type of model is often called an overdispersed Poisson model, which is functionally similar to a negative binomial model (see Rabe-Hesketh & Skrondal, 2008, pp. 389–390, or Ver Hoef & Boveng, 2007, for further discussion). Moreover, adding the extra error term is very similar to a residual error term, as in models for normally distributed data, and can be assessed via a deviance test. Model 2 in Table 1 fits an overdispersed Poisson GLMM, incorporating this extra error term. A deviance test for assessing the improvement in fit due to the addition of the per-observation error term is highly significant, χ²(1) = 2,032.8, p < .01.

Earlier, we commented on the number of zeroes descriptively in the RAPI data, and a final component with count regression models is to examine how well the model replicates the number of zeroes, and more generally, the distribution of counts in the raw data; that is, what does the model predict the histogram of counts looks like, and specifically, what the zeroes look like? A figure included with the supplementary online material presents the raw distribution of the RAPI along with estimates from model two in Table 1 (see Figure 2 at http://dx.doi.org.offcampus.lib.washington.edu/10.1037/a0029508.supp). Although Model 2 appears quite accurate for counts greater than 2, there is noticeable lack of fit for counts of 0, 1, and 2. Essentially, the Poisson distribution is not flexible enough to fit the steep drop of counts from 0 to 1, with much more moderate change for counts greater than 1. As a result the raw RAPI data has 756 zeroes and 395 ones, whereas the final overdispersed Poisson GLMM predicts 616 zeroes and 546 ones. This is a primary statistical motivation for considering a zero-inflated or hurdle mixed model.

Following our example with the RAPI, a hurdle mixed model would be     where t indexes time, i indexes individuals, l and c index logistic and count portions of the model, and p is the proportion of RAPI scores greater than 0. As noted earlier, zero-inflated and hurdle models have two submodels, one related to the zeroes and a second related to the counts. The key difference between hurdle and zero-inflated models is how they handle zeroes: Hurdle models cleanly divide the models, with all zeroes accounted for in the logistic regression, whereas zero-inflated models treat the zeroes as a mixture (i.e., both submodels in zero-inflated models contribute zeroes). As we saw with the overdispersed Poisson GLMM, the hurdle GLMM presented above adds random effects to the linear predictors, with the major difference now being that there are two linear predictors. The random intercept in the logistic model implies that there is variability across individuals in the likelihood of reporting any problems, whereas the random intercept and slope in the count regression models variability in the intercept and change across time in number of problems when there are problems reported. A random slope for time in the logistic portion would model individual variability over time in proportion of zeroes.

Prior to fitting the hurdle mixed model, let us say a word about when to use zero-inflated or hurdle models. Similar to many model selection decisions, the choice between models should include statistical considerations, theoretical considerations, and parsimony. As noted earlier the overdispersed Poisson GLMM of the RAPI data underpredicted zeroes (756 in the raw data vs. 616 predicted by the model), and with the TLFB data, the same type of model led to a highly non-normal distribution for the per-observation error term, based on a histogram and quantile-quantile plot of this error term. Thus, in both of these cases a zero-inflated or hurdle mixed model might be preferred on statistical grounds. However, the statistical motivations just noted do not automatically mean that covariates have important and distinct relationships across the logistic and count portions of a hurdle (or zero-inflated) model. That is, sometimes a hurdle or zero-inflated mixed model will fit the data better, but the conclusions are largely the same compared to a Poisson GLMM (typically because the covariates are primarily related to the nonzero counts). In this scenario, parsimony might suggest the simpler Poisson GLMM is adequate as it yields similar conclusions, though we strongly suggest fitting the more complex model to test whether the simpler model is adequate. Finally, theory may make clear predictions about the proportion of days with any drinking or the amount drunk on days with drinking. These would clearly prefer hurdle or zero-inflated models.

Results from the hurdle mixed model shown in Equations 2a and 2b are presented in Table 2. In examining Table 2, we find that these results map on to the earlier descriptive graphs quite closely. In particular, the count portion of the model shows that men tend to have more alcohol problems when there are problems at the start of the study (RR = 1.21). Because of the coding of the gender dummy-variable, the main effect of time describes the change in women's alcohol problems across time. The RR of 0.98 implies that, conditional on the random effects, women's RAPI (i.e., alcohol problems when there are problems) is dropping 2% with each successive assessment, whereas the interaction between gender and time implies that men's alcohol problems (when there are problems) are barely changing at all, which is confirmed by estimating the men's simple slope for time, RR = 0.99, 95% CI [0.98, 1.00].

Over-Dispersed Poisson Hurdle Mixed Model Results for RAPI Data

The logistic portion of the model describes the proportion of the sample reporting some alcohol problems. (Note that software for zero-inflated and hurdle models are not always consistent as to whether the logistic portion predicts zeroes as opposed to nonzeroes. Here it is predicting nonzeroes.) The ORs in Table 2 show that there are no differences between the sexes in the proportion of individuals reporting alcohol problems at the start of the study, OR = 0.83, 95% CI [0.59, 1.18]. For women, the odds of reporting any alcohol problems decreases by 3% per month, OR = 0.97, 95% CI [0.95, 0.99], conditional on the random effects. For men, their rate of change is somewhat slower, OR = 1.02, 95% CI [0.99, 1.04].

In turning to the TLFB data, the earlier plots showed that there was a very large stack of zeroes and that covariate effects (i.e., gender, weekend, and fraternity/sorority status) might vary across submodels defined by likelihood of any drinking (i.e., zero vs. not zero) and amount of drinking when there is drinking. These considerations would point toward a zero-inflated or hurdle model; moreover, preliminary models using an overdispersed Poisson GLMM showed that there was a very poor fit between predicted and observed count distribution, and the per-observation random-effect distribution was not close to being normally distributed. Given these considerations, a hurdle overdispersed Poisson GLMM was fit to the TLFB data, fitting main and two-way interaction effects in both submodels via three dummy variables: Male (0 = Women, 1 = Men), Weekend (0 = Sun/Wed, 1 = Thurs/Sat), and FratSor (0 = Not in fraternity/sorority, 1 = Fraternity/Sorority). A random intercept and random slope for weekend were included in both the logistic and truncated count submodels, as well as the per-observation random intercept in the truncated count submodel. Conditional (or unit-specific) fixed-effects are shown in Table 3. In the logistic portion of the model, there is a single, significant effect for weekend: Many more students drink, regardless the quantity, on the weekends as compared to the weekdays, OR = 3.06, 95% CI [2.60, 3.54]. In the count submodel, all three two-way interactions are significant. One approach to interpreting interactions is to use “simple slopes” (see, e.g., Jaccard, 2001). However, in the present case, the model is estimating mean drinking on drinking days for eight specific groups—for each combination of gender, weekend, and fraternity/sorority status. To help interpret the findings of the count submodel, we estimated marginal (or population-average), predicted means for each of these eight groups, which are shown in Figure 7. The figure reveals several interesting findings. Not surprisingly, fraternity and sorority members drink more on drinking days than students who are not in fraternities and sororities, and fraternity and sorority members' drinking is similar on weekdays versus weekend—when they are drinking (i.e., the count submodel in a hurdle is only for nonzeroes). However, contrasts of weekend versus weekday drinking for men and women who are not in fraternities and sororities shows that they do drink more on weekends, women: RR = 1.07, 95% CI [0.99, 1.14], p = .06; men: RR = 1.20, 95% CI [1.12, 1.28], p < .01. Thus, the hurdle mixed model provides a more nuanced view of drinking differences across these groups, parsing out drinking versus not drinking and mean drinking on drinking days.

Over-Dispersed Poisson Hurdle Mixed Model Results for TLFB Data

Figure 7. Predicted (marginal) drinks for each unique combination of gender, weekend (WE) versus. weekday (WD), and fraternity/sorority status from count submodel of overdispersed Poisson hurdle mixed model.

Statistical software packages are steadily expanding their coverage of GLMMs and the models covered in the current paper. Table 4 presents an overview of SPSS, Stata, R, SAS, and Mplus, describing which of the models covered in the present article are possible to fit. As seen there, the basic Poisson GLMM is implemented in all software, whereas negative binomial GLMM and zero-altered models are less widely available. The online materials provide computer code for fitting some example models in each of the packages mentioned, and the online supplementary material covers additional details related to fitting GLMM, which is both more technical but still practically important (see http://dx.doi.org.offcampus.lib.washington.edu/10.1037/a0029508.supp). Not all software packages were able to fit all models described in the present article (e.g., SAS NLMIXED can fit zero-inflated mixed models but could not fit all the variance components describe in the TLFB example). Over the coming years, there will likely be greater software availability for fitting the types of models covered here as well as more robust fitting algorithms.

Comparison of Software for Fitting Generalized Linear Mixed Models (GLMM)

Summary

The present article discussed extensions to count regression and zero-altered count regression models to longitudinal data based on GLMM. We hope that this presentation, along with the appendix and available data and code, helps addiction researchers to learn and appropriately apply these models. Given the research designs and data that addiction researchers often collect, the models covered here are often the most appropriate analytic tools. At the same time, like other statistical models that are finding their way to the applied research literature (e.g., growth mixture models; Muthén & Muthén, 2000), multilevel count models are complex. The types of assumptions and general considerations in model fitting grow exponentially from linear models to generalized linear models to linear mixed models to generalized linear mixed models. We encourage researchers to explore and use these tools as their hypotheses and data merit, though also note that they require a serious commitment of time and study to be used appropriately.

¹ The six graphs in Figure 4 were created by simulating 1,000 random draws from the specified Poisson or negative binomial distributions. Thus, these are similar to frequency histograms, except there is no binning of multiple values.

² There are alternatives to the negative binomial distribution in terms of dealing with overdispersed count data, including quasi-Poisson models, Poisson-normal models, and robust standard errors. Details on these models can be found in Hilbe (2011).

³ Note that the distribution of slopes appears somewhat odd, with spikes at extremely high or low values. This reflects that some individuals have incomplete data, and hence, slopes fit to two or three points of data can yield very extreme values. The subject-specific predictions from GLMMs pool information about the sample and the individual's data, leading to more sensible fits for individuals (see, e.g., discussion of empirical Bayes estimates in Raudenbush & Bryk, 2002).

⁴ The notation in Equation 1 is slightly nonstandard. In the statistical literature it is common to see GLMMs described in terms such as: η = XB + Zb, with g(μ) = η, and g( ) = log. Translating into words this means that η is the linear predictor, or right-hand side of the model, including both fixed and random effects (i.e., XB and Zb, respectively). The linear predictor is connected to the mean of the outcome (i.e., μ) via some function. With the Poisson model, the link function is the log.

⁵ Estimating the predicted number of zeroes based on a GLMM takes us into somewhat more technical material. The Poisson GLMM can be represented by Pois(XB + Zb), where X is the design matrix of the fixed-effects; B is a vector of fixed-effects coefficients; Z is the design matrix of the random effects; and b is a vector of random-effects variances. The key thing to realize is that the formula within the parentheses defines the fitted values of the model. Thus, we can read this equation as follows: The data are distributed as a Poisson random variable with a mean structure defined by the fixed and random effects of the GLMM. Similar to the discussion of conditional effects, the random effects create additional challenges in estimating the model-based distribution of counts. The accompanying R code shows one method for simulating from the fitted model to estimate the predicted distribution of counts.

⁶ We note in passing that the hurdle mixed models reported in this article were fit using the MCMCglmm package (Hadfield, 2010) in R, which uses a Bayesian approach to model estimation. For many practical problems maximum likelihood estimation and Bayesian MCMC estimation will yield similar if not identical results, though there are basic differences in both inferential philosophy as well as estimation. A brief discussion is provided in the technical appendix of the online material (see http://dx.doi.org.offcampus.lib.washington.edu/10.1037/a0029508.supp). In this particular instance, the choice is quite pragmatic as MCMCglmm provides the most flexible option for fitting these types of models in R.

⁷ To estimate marginal (or population-average) estimates from a GLMM, it is necessary to include both fixed and random effects. The online supplementary material provides details and the example of how the present estimates were created (see http://dx.doi.org.offcampus.lib.washington.edu/10.1037/a0029508.supp).

Atkins, D. C., & Gallop, R. J. (2007). Rethinking how family researchers model infrequent outcomes: A tutorial on count regression and zero-inflated models. Journal of Family Psychology, 21, 726–735. doi:10.1037/0893-3200.21.4.726

Bolker, B. M., Brooks, M. E., Clark, C. J., Geange, S. W., Poulsen, J. R., Stevens, M. H. H., & White, J. S. (2009). Generalized linear mixed models: A practical guide for ecology and evolution. Trends in Ecology and Evolution, 24, 127–135. doi:10.1016/j.tree.2008.10.008

Breslow, N. E., & Clayton, D. G. (1993). Approximate inference in generalized linear mixed models. Journal of the American Statistical Association, 88, 9–25. doi:10.2307/2290687

Buntin, M. B., & Zaslavsky, A. M. (2004). Too much ado about two-part models and transformation? Comparing methods of modeling Medicare expenditures. Journal of Health Economics, 23, 525–542. doi:10.1016/j.jhealeco.2003.10.005

Coxe, S., West, S. G., & Aiken, L. S. (2009). The analysis of count data: A gentle introduction to Poisson regression and its alternatives. Journal of Personality Assessment, 91, 121–136. doi:10.1080/00223890802634175

Draper, D. (2008). Bayesian multilevel analysis and MCMC. In J.de Leeuw and E.Meijer (Eds.), Handbook of multilevel analysis (pp. 77–139). New York, NY: Springer. doi:10.1007/978-0-387-73186-5_2

Gaher, R. M., & Simons, J. S. (2007). Evaluations and expectancies of alcohol and marijuana problems among college students. Psychology of Addictive Behaviors, 21, 545–554. doi:10.1037/0893-164X.21.4.545

Gelman, A., & Hill, J. (2007). Data analysis using regression and multilevel/hierarchical models. New York, NY: Cambridge University Press.

Hadfield, J. D. (2010). MCMC methods for multi-response generalized linear mixed models: The MCMCglmm R package. Journal of Statistical Software, 33, 1–22.

Heagerty, P. J., & Zeger, S. L. (2000). Marginalized multilevel models and likelihood inference. Statistical Sciences, 15, 1–26. doi:10.1214/ss/1009212670

Hedeker, D., & Gibbons, R. D. (2006). Longitudinal data analysis. Hoboken, NJ: John Wiley & Sons.

Hilbe, J. M. (2011). Negative binomial regression (2nd ed.). New York, NY: Cambridge University Press. doi:10.1017/CBO9780511973420

Jaccard, J. (2001). Interaction effects in logistic regression. Newbury Park, CA: Sage.

Lewis, M. A., Neighbors, C., Geisner, I. M., Lee, C. M., Kilmer, J. R., & Atkins, D. C. (2010). Examining the associations among severity of injunctive drinking norms, alcohol consumption, and alcohol-related negative consequences: The moderating roles of alcohol consumption and identity. Psychology of Addictive Behaviors, 24, 177–189. doi:10.1037/a0018302

Liang, K. Y., & Zeger, S. L. (1986). Longitudinal data analysis using generalized linear models. Biometrika, 73, 13–22. doi:10.1093/biomet/73.1.13

Lynch, S. M. (2007). Introduction to applied Bayesian statistics and estimation for social scientists. New York, NY: Springer. doi:10.1007/978-0-387-71265-9

McCullagh, P., & Nelder, J. A. (1989). Generalized linear models (2nd ed.). Boca Raton, FL: Cahpman & Hall/CRC.

Molenberghs, G., & Verbeke, G. (2005). Models for discrete longitudinal data. New York, NY: Springer.

Muthén, B., & Muthén, L. (2000). Integrating person-centered and variable-centered analysis: Growth mixture modeling with latent trajectory classes. Alcoholism: Clinical and Experimental Research, 24, 882–891. doi:0145-6008/00/2406-0882

Neal, D. J., & Simons, J. S. (2007). Inference in regression models of heavily skewed alcohol use data: A comparison of ordinary least squares, generalized linear models, and bootstrap resampling. Psychology of Addictive Behaviors, 21, 441–452. doi:10.1037/0893-164X.21.4.441

Neighbors, C., Atkins, D. C., Lewis, M. A., Lee, C. M., Kaysen, D., Mittmann, A., . . .Rodriguez, L. M. (2011). Event specific drinking among college students. Psychology of Addictive Behaviors, 25, 702–707. doi:10.1037/a0024051

Neighbors, C., Lewis, M. A., Atkins, D. C., Jensen, M. M., Walter, T., Fossos, N., . . .Larrimer, M. E. (2010). Efficacy of web-based personalized normative feedback: A two-year randomized controlled trial. Journal of Consulting and Clinical Psychology, 78, 898–911. doi:10.1037/a0020766

Rabe-Hesketh, S., & Skrondal, A. (2008). Multilevel and longitudinal modeling using Stata. College Station, TX: Stata Press.

Raudenbush, S. W., & Bryk, A. S. (2002). Hierarchical linear models: Applications and data analysis methods (2nd ed.). Thousand Oaks, CA: Sage Publications.

Raudenbush, S. W. (2008). Many small groups. In J.de Leeuw & E.Meijer (Eds.), Handbook of multilevel analysis (pp. 207–236). New York, NY: Springer.

Rodriguez, G., & Goldman, N. (2001). Improved estimation procedures for multilevel models with binary response: A case-study. Journal of the Royal Statistics Society, Series A, 164, 339–355.

Simons, J. S., Neal, D. J., & Gaher, R. M. (2006). Risk for marijuana-related problems among college students: An application of zero-inflated negative binomial regression. American Journal of Drug and Alcohol Abuse, 32, 41–53. doi:10.1080/00952990500328539

Singer, J. D., & Willett, J. B. (2003). Applied longitudinal data analysis: Modeling change and event occurrence. New York, NY: Oxford University Press.

Snijders, T. A. B., & Bosker, R. J. (1999). An introduction to basic and advanced multilevel modeling. Thousand Oaks, CA: Sage Publications.

Ver Hoef, J. M., & Boveng, P. L. (2007). Quasi-Poisson vs. negative binomial regression: How should we model overdispersed count data?Ecology, 88, 2766–2772. doi:10.1890/07-0043.1

White, H. R., & Labouvie, E. W. (1989). Towards the assessment of adolescent problem drinking. Journal of Studies on Alcohol, 50, 30–37.

Zeileis, A., Kleiber, C., & Jackman, S. (2008). Regression models for count data in R. Journal of Statistical Software, 27, 1–25.

Submitted: November 22, 2011 Revised: June 15, 2012 Accepted: June 18, 2012

Correction to Atkins et al. (2012)

In the article “A Tutorial on Count Regression and Zero-Altered Count Models for Longitudinal Substance Use Data” by David C. Atkins, Scott A. Baldwin, Cheng Zheng, Robert J. Gallop, and Clayton Neighbors (Psychology of Addictive Behaviors, Vol. 27, No. 1, pp. 166–177. doi:10.1037/a0029508), the URL for the supplemental material was incorrect throughout the text due to a production error.

Supplemental material for this article is available at: http://dx.doi.org.offcampus.lib.washington.edu/10.1037/a0029508.supp. The online version of this article has been corrected.

Adolescent Change Language Within a Brief Motivational Intervention and Substance Use Outcomes

Acknowledgement: The research reported herein was supported by National Institute on Drug Abuse Grant R01 DA15751. We gratefully acknowledge the contributions of research assistants Sarah Bowen and Dana Rhule; counselors Jennifer Mullane, Melissa Phares, and Maija Ryan; data coders Avry Todd and Kate Hallman; the staff at New Horizons Ministries; and Dan Kivlahan for comments on a draft of this article.

Motivational interviewing (MI; Miller & Rollnick, 2002) is a popular, empirically based counseling method for a range of health-related problems. Defined as a “directive, client-centered counseling style for eliciting behaviour change by helping clients explore and resolve ambivalence” (Rollnick & Miller, 1995, p. 326), MI emphasizes formation of collaborative therapeutic relationships with clients through which client language about change may be strategically elicited and reinforced. Change language is defined as client expressions of problems with the current state, benefits of change, and hope and optimism about future change (Miller & Rollnick, 2002). In articulating a theory for MI, Hettema, Steele, and Miller (2005) stated that (a) the practice of MI should elicit increased levels of change and decreased levels of resistance from clients, (b) the extent to which clients verbalize arguments against change (resistance) during MI will be inversely related to the degree of subsequent behavior change, and (c) the extent to which clients verbalize change talk (arguments for change) during MI will be directly related to the degree of subsequent behavior change.

MI should be considered among therapeutic approaches based in the “ordinary language” of clients, in that therapeutic discourse is understood on the basis of common interpretations of what clients say about thoughts, feelings, and behavior (Moyers et al., 2007). The empirical relationship between client language and outcomes thus assumes a central role in the suggested impact of MI; yet evidence for this hypothesized relationship is limited. With respect to verbalizations contrary to change, Miller, Benefield, and Tonigan (1993) reported that independent ratings of client interrupting, arguing, off-task responses, and other negative responses within a brief intervention about alcohol use were strongly predictive of poorer 1-year drinking outcomes. Recently, Moyers et al. (2007) coded client language of participants randomly selected from each of three Project MATCH therapy conditions. Rates of both positive and counter–change language in the first clinical session across treatments predicted follow-up drinking outcomes.

Further evidence of a link between positive client change language and clinical outcomes in MI sessions was reported by Amrhein, Miller, Yahne, Palmer, and Fulcher (2003) in sessions with illicit drug users as they presented for treatment. Client language was subjected to a linguistic analysis that subdivided change language into elements thought to vary with respect to the nature and degree of commitment to change (commitment, desire, ability, need, readiness, and reasons). Clients who reduced their substance use during and after treatment made more frequent commitment to change statements during the evaluation of a change plan at the end of the interview compared with those who intermittently used substances over time or who never improved. Other aspects of change language were predictive of commitment language.

This small, emerging evidence base provides some support for the notion that client change language can prospectively predict clinical outcomes. These studies, however, have been completed with adults who were seeking treatment (or a free checkup; Miller et al., 1993). Yet MI has been adopted in numerous clinical and nonclinical contexts (Miller & Rollnick, 2002). One common focus for brief interventions in opportunistic contexts is with young people, who seldom seek services on the basis of their own concerns for health and safety (Baer & Peterson, 2002; Monti, Colby, & O'Leary, 2001). Despite this interest, we are unaware of studies that describe the target of MI sessions, describe youth verbalizations about behavior change, or examine whether these relate to current or future behavior. Such information would seem important given that aspects of adult language are still developing in adolescents (Nippold, 2000; O'Kearney & Dadds, 2005). Hence, adolescents receiving feedback about drug use risks may talk about change very differently than adults seeking treatment.

The current analyses sought to extend prior studies of client language within MI sessions by examining the language of homeless youths who use alcohol and illicit substances and were recruited for a risk reduction program (Baer, Garrett, Beadnell, Wells, & Peterson, 2007) that utilized a brief motivational intervention (BMI). We tested whether global ratings of client behavior in MI sessions, and behavioral ratings of positive and counter–change language and commitment talk, would predict changes in substance use from baseline to 1- and 3-month follow-ups.

Design and Procedures

All procedures were approved by the University of Washington Institutional Review Board. A sample of youths (ages 13–19 years) was recruited from a nonprofit, faith-based drop-in center (Baer et al., 2007) by study counselors. Inclusion criteria included lack of stable housing and report of substance use in the prior 30 days. The 127 youths were randomly assigned to receive four sessions of a BMI (n = 75) or drop-in center treatment-as-usual (n = 52), which included no specific clinical intervention, although case management and other services were available. Only youths receiving BMI are included in current analyses. As MI was not found to be superior to treatment-as-usual in this study (see Baer et al., 2007), and treatment-as-usual received no specific treatment sessions, mediating tests of the function of client change for MI were not possible. Follow-up assessments were conducted via appointments and intercept at the drop-in center and on the streets both 1 month and 3 months postbaseline. Following baseline interviews, youths assigned to BMI stayed for the first of four BMI sessions, with later BMI sessions scheduled in the following 4 weeks. Baseline interviews and the BMI sessions were conducted by a master's-level clinician, with follow-up interviews conducted by alternative project staff. As described elsewhere (Baer et al., 2007), participants received cash for completing study assessments and vouchers redeemable at local retail stores for completed BMI sessions. With consent of participants, baseline interviews and BMI sessions were audiotaped.

Sample description and retention

In the original study attrition was not associated with experimental condition, baseline demographic, or substance use rates. Participants identifying themselves as minority racial group members were retained at slightly higher rates compared with those self-identified as nonminority. Of 75 youths assigned to BMI, 21 were removed from analyses due to extensive missing data or concerns about validity of reported substance use. These included youths who (a) were incarcerated in the 30 days prior to baseline assessment (n = 3) or a follow-up assessment (n = 4), (b) received strongly negative interviewer ratings regarding consistency of their responses (n = 2), (c) refused to be audiotaped (n = 6) or were eliminated because of taping malfunction for at least half of the session (n = 4), (d) attended only highly abbreviated sessions (n = 1), and (e) missed both follow-up assessments (n = 1). No differences were seen between those included versus those excluded in relation to gender, age, or racial or ethnic minority (all ns). At baseline the 21 excluded by these criteria reported less substance use in the previous 30 days than the 54 included (days abstinent M = 14.0, SD = 8.8 vs. M = 7.5, SD = 8.9), t(73) = 2.84, p < .01.

The resulting youth sample for current analysis showed a fairly even gender distribution (54% male vs. 46% female) and mean age of 17.9 years (SD = 1.3). Reported ethnicity was 59% Caucasian, 17% multiracial, 7% Native American, 7% African American, 6% Hispanic or Latino, and 4% Asian or Pacific Islander. The sample reported high rates of substance use, with a mean of 7.5 (SD = 8.9) days of abstinence from substances in the month prior to baseline assessment. Alcohol use in the past month was reported by 89.1% of the sample. Marijuana was the most commonly used drug in the past 30 days by youth report (94.4%), followed by “club drugs” (57.1%), methamphetamine (53.5%), hallucinogens (35.9%), cocaine (33.3%), and opiates other than heroin (33.3%).

Intervention description

The BMI followed the model of an extended substance use “checkup” in which information or exercises about patterns and risks of substance use are provided as personal feedback (Miller, Sovereign, & Krege, 1988). Intervention content is described elsewhere (Baer et al., 2007). Interventionists were trained in MI techniques, then supervised via regular-session audiotape review by John S. Baer (a licensed psychologist, member of the Motivational Interviewing Network of Trainers, and an experienced MI trainer). Interventionists were trained to be nonconfrontational in tone and to provide advice about risk reduction only with permission. Initial sessions averaged 17 min (SD = 7), with subsequent sessions averaging 33 min (SD = 11). Of the 54 youths assigned to BMI, 27 attended all four sessions, 9 attended three, 9 attended two, and 9 attended one. Mean intervention time across sessions was 79.8 min (SD = 42.9). Youth ratings of the intervention were overwhelmingly positive (Baer et al., 2007).

Measures

Substance use frequency and severity

Participants were asked to recall substance use in the prior 30 days through a modified Time Line Follow-Back interview (Sobell & Sobell, 1993). Given that individuals used many combinations of substances over the 30-day period, an omnibus measure of “days of abstinence” from alcohol, marijuana, and other drug use (excluding tobacco) was calculated. We used abstinence days for conceptual clarity because days of use may represent days of use of any number of substances, whereas abstinence is clearly defined by no substance use.

Validation of drug use self-report

Urinalysis was used to calculate sensitivity and specificity of self-reported cocaine, amphetamine, opiate, and marijuana use at the 3-month follow-up. With 41 of 54 youths (76%) providing urinalysis, no evidence of systematic underreporting was observed (Baer et al., 2007).

Indices of intervention discourse

Audiotaped BMI sessions were coded by three trained raters. The scoring system utilized components of the Motivational Interviewing Skill Code (Version 1.0; Miller, 2000; Moyers, Martin, Catley, Harris, & Ahluwalia, 2003) and psycholinguistic framework of Amrhein et al. (2003) to characterize adolescent language in BMI sessions. Based on the Motivational Interviewing Skill Code, the scoring system included three global ratings of client qualities on 7-point Likert scales (1 = not at all, 7 = strong): (a) affect, how openly and directly emotions were expressed; (b) cooperation, how much responsibility for completing in-session tasks was shared; and (c) disclosure, how much personal information was revealed. An additional global score, task orientation, was added to reflect how much the adolescent remained focused and engaged during therapeutic tasks.

In line with Amrhein et al. (2003), frequency of youth statements about change, termed change talk (Miller & Rollnick, 2002), was tallied relative to substance-related risk reduction, promotion of health and safety, or service utilization. Three categories were coded to reflect differing types of language and degree of interest in change noted in prior research (Amrhein et al., 2003) and based on qualities of adolescent language observed in this and previous studies. These included (a) commitment, or statements of explicit intention to change (or maintain) behavior (e.g., “I'm not going to use meth again” or “I'm never going to stop smoking pot”); (b) reasons, or statements providing a rationale for (or against) behavior change (e.g., “When I'm high I fight with my girlfriend” or “Using drugs helps me cope with being on the street”); and (c) desire/ability, or statements that indicate desire, willingness, or ability or self-efficacy to change or not to change (e.g., “I'd like to cut back my drinking” or “I can't deal with not using because I know it'll be too painful”). Irrespective of categorization, each instance of change talk was also assigned a positive or negative valence reflecting if the statement was in favor of or opposed to reduced substance use, reduced health risks, or increased service utilization. Coders recorded tallies of each type of change talk across 5-min intervals within each session. To control for varying degrees of talkativeness, session lengths, and attendance, we converted tallies to average rates of change talk per 5 min. Overall rates of each type of change talk were calculated across individual sessions, or across all sessions attended, as the total tally across 5-min intervals divided by number of 5-min intervals.

Training and Reliability of Raters

Three raters received extensive training in recognizing and rating MI elements in therapeutic dialogue and attended weekly supervision meetings where scoring dilemmas were discussed. Interrater reliability was assessed from a subset of sessions that all three rated (n = 15, 8.3%). Intraclass correlations, computed as a two-way random model, were acceptable on the basis of Cicchetti (1994) criteria (above .40), although lowest for commitment language (see Table 1).

Coding Reliability, Means, and Standard Deviation for Youth and Counselor Language, Averaged Across Sessions

Descriptives for session codes are shown in Table 1. For the 7-point scale, mean global scores for cooperation, affect, and task orientation were between 4 and 5, whereas youth disclosure was just above 5. Youths more often expressed reasons for, reasons against, and desire/ability for change, and less often expressed desire/ability against, commitment for, and commitment against change.

Intercorrelations between global scores, change language, and substance use rates at baseline and follow-up assessments are presented in Table 2. As can be seen, desire/ability comments against change were significantly correlated with substance use rates at follow-up assessments. There also are several modest correlations among global ratings and among rates of use of different forms of change language. Among desire/ability and commitment, respectively, statements for and against change were correlated. This may reflect greater talkativeness or suggest that youths may tend to use one form of change language to express both positive and negative motivation. Partial correlations (controlling for the sum of other talk types) suggest that the former may be true for desire/ability for and against change (r = .35, p < .05 vs. partial r = .23, ns) and the latter true for commitment for and against change (r = .40, p < .01 vs. partial r = .39, p < .01).

Intercorrelations Among Language Codes, Global Ratings, and Substance Use Rates (n = 50–54)

Multiple regression assessed whether session language predicted changes in substance use rates. For each follow-up point, multiple regression predicting substance use abstinence in the prior 30 days was performed in Mplus 5.0 through robust maximum likelihood estimation, for sets of predictors. Three participants were missing assessment at 1 month and 4 at 3 months (all 54 had data for at least one follow-up). To assess prediction of change in substance use rates and control for treatment exposure, we used regressions that included baseline prior 30-day abstinence and the number of feedback sessions attended. Demographic characteristics (gender, race, age) were unrelated to either predictors or outcomes in preliminary analyses and were excluded in regressions to preserve statistical power in this small sample. Because counselors were not associated with substance use outcomes in the original study, they also were excluded in regression analyses. As noted in our prior study (Baer et al., 2007), days of abstinence increased over time for those receiving BMI (not differing from those in the control group). For youths included in current analyses, average days of abstinence over the prior 30 days was 7.5 (SD = 8.9) at baseline, then increased to 11.5 (SD = 10.8) at 1-month follow-up and 11.0 (SD = 10.6) at 3-month follow-up.

Tests of the four youth global scores on change in days of abstinence showed that higher task orientation predicted more days of abstinence at the 3-month but not the 1-month follow-up (standardized β = .44, p < .01). No other effects were statistically significant.

In regressions predicting 1-month abstinence rates from youth verbalizations about change, two of the six youth talk rates had statistically significant effects. Reasons in favor of change predicted greater abstinence, and desire/ability against change predicted less abstinence (standardized β = .23 and −.57, respectively, p < .05 and .001). Only desire/ability against change predicted less abstinence at the 3-month time point (standardized β = −.41, p < .05). Figure 1 illustrates the magnitude of effect based on desire/ability against change. Those who expressed more than the median desire/ability against change reported mean abstinence days at the 1- and 3-month follow-ups of 6.9 and 8.7 (SD = 9.1 and 10.0, respectively), whereas those making less than the median amount reported mean abstinence days of 16.2 and 13.4 (SD = 10.5 and 11.0, respectively).

Figure 1. Days of abstinence depending on desire/ability against change talk (individuals above vs. below the median).

Supplemental analyses addressed possible alternative interpretations of results from regression analyses. Outlier analyses (Cohen, Cohen, West, & Aiken, 2003) in which Cook's distance and leverage scores were computed and outliers were removed showed no substantive changes in these findings. Regressions for 1- and 3-month outcomes including significant predictors from the two original regression analyses (global ratings and rates of change language) suggested that the effect of task orientation on 3-month abstinence did not persist once change language was included. Additional analyses revealed no predictive effects of youth talk in the first BMI session, of changes in rates of language across sessions, or differential effects of change talk rated in the first half versus second half of sessions.

Finally, we explored why reasons for change had a nonsignificant zero-order correlation with 1-month abstinence (r = −.03; see Table 2) but was a significant predictor in multivariate regressions (including those controlling for outlier effects). Review of partial correlations with other variables in the regression model suggested statistical suppression only in relation to desire/ability against change (partial r = .27, p = .065). A 2 × 2 analysis of variance of the relation between these two variables in prediction of change in substance use rates, on the basis of median split of each, suggested an interaction, F(1, 3) = 3.10, p = .09. When desire/ability against change was high, there was a detrimental effect on change in abstinence regardless of whether reasons for change was low or high (mean abstinence change = 1.0 and 0.5, for those with low and high reasons for change, respectively, SD = 5.5 and 7.9); however, when desire/ability against change was low, reasons for change appears to affect abstinence (mean abstinence change = 2.9 and 11.7, for those with low and high reasons for change, respectively, SD = 11.2 and 10.3).

Our examination of youth verbalizations during BMI sessions provides support for one of the basic tenets of MI: that client change language is related to subsequent behavior change. Despite myriad psychological and social problems among homeless youths and their general disengagement from broader social systems, and despite that this sample was not seeking treatment, change language in BMI sessions regarding substance use or service utilization significantly and prospectively predicted changes in substance use. Analyses suggest that two aspects of youth language about change are differentially indicative of actual behavioral change: statements about reasons in favor of change and statements counter to the desire or ability to change.

The strongest effect was noted for desire/ability language against change. Such comments were strongly and prospectively associated with substance use reductions at both follow-ups, despite being fairly uncommon (mean frequency of .61 per 5 min). Thus, data suggest that within BMI sessions with adolescents, a few comments about not wanting, needing, or being able to change bode poorly for subsequent reduction in substance use. Such utterances may be a marker of a specific type of resistance in MI sessions; resistance has been associated with poor clinical outcomes among adolescents (Karver, Handelsman, Fields, & Bickman, 2006; Shirk & Karver, 2003). This finding is also consistent with literature on behavioral intentions (Fishbein & Ajzen, 1975; Fishbein, Hennessy, Yzer, & Douglas, 2003; Webb, Baer, Getz, & McKelvey, 1996), despite these observational codes differing from typical measures of attitudes and behavioral intention.

Stated reasons for change also reveal a significant prospective relation with outcomes, albeit smaller than that noted for comments about desire/ability against change and possibly relevant only when statements of desire/ability against change are absent. Prior research (Baer, Peterson, & Wells, 2004) suggests that most homeless youths describe themselves as precontemplative or contemplative about change. We expected that, at best, homeless youths would express ambivalence about change in substance use. Given the brief intervention process where feedback is provided about risks and counselors strategically work to elicit positive reasons for change, we expected that problem recognition expressed through reasons for change might emerge. Compared with other forms of change language, positive reasons for change were the most common form of verbalization (1.32 per 5-min interval). Still, this rate does not suggest that these BMI sessions were filled with discussions about negative aspects of substance use and benefits of reduction and cessation. A relatively few stated reasons for change may be important for future behavior nonetheless, especially if youths are not otherwise negative with respect to their desire or ability to change. Our data do not suggest that the rate of such language changed across or within sessions.

Commitment language, which Amrhein et al. (2003) have argued as most important in the prediction of substance use outcomes, was not found to be associated with change in this study. Amrhein et al.'s study differed from this one in that it took place among adults who were seeking treatment for drug problems, and the brief MI intervention included a discussion of a change plan at the end. In contrast, our protocol with homeless youths varied from one to four sessions and followed a booklet from which adolescents could choose topics for discussion. In this context, commitment language was rare (0.27 per 5 min in favor of change, 0.18 per 5 min against change), perhaps accurately reflecting the fact that youths were not seeking assistance. Yet an intervention that includes exercises that elicit commitment language (such as developing a change plan) might produce positive findings about commitment language. Additionally, commitment language among these youths proved difficult to code; intraclass correlations were lowest for this category. A different rating method for this language category might lead to different results.

It is important to acknowledge that this study tests the relationship between youth verbalizations and subsequent behavior only and does not test a complete causal chain for MI efficacy. Several additional features of the study should cause readers to draw conclusions cautiously. We would be more confident in these relationships if we had a larger sample and more variability in youth substance use. Despite a careful approach to data analysis (including evaluating the influence of few cases or outliers), the inherent overfitting of regression models requires that the relationships we have reported be replicated. The study is also limited by the use of self-reported substance use (although urine testing suggests that reports were not systematically biased). Results are based on those youths who were not recently incarcerated and who allowed audiotaping. Generalization to other adolescent populations, both housed and homeless, requires study. Results also relied on a specific coding scheme that was informed by prior studies yet altered by investigators for this study. Alternative coding schemes, which either combine or split constructs differently, may suggest different patterns and predictive relationships. Although we analyzed only those codes that reached a minimum level of confidence based on Cicchetti's (1994) standards, higher and more consistent interrater reliability would enhance confidence in the observed relationships.

Although not providing a direct mediating test, the relationships reported here should provide support for those who design brief interventions in which counselors are taught to pay careful attention to and elicit change language from adolescent clients. Youth verbalizations about change may provide one method of assessment for likelihood of change, which could lead to tailored interventions. For the clinician, taping, transcribing, and reliably coding verbalizations are potentially unfeasible. If these results are replicated, attention to more parsimonious means of assessing change talk will allow for clinical application.

Our data suggest that negative comments about desire or ability against change, even when observed infrequently, are relatively strongly predictive of poor outcomes. Additionally, even among youths with multiple social and psychological difficulties, positive statements about reasons for change may indicate that risk reduction is likely. Whether and how the process of language development itself moderates this relationship would be a fruitful area for exploration. In addition, it is left for future research to establish that the specific eliciting and reinforcing of such change language (e.g., through MI) can in fact alter the course of substance use and other risk behavior among high-risk youths.

¹ To assess beliefs about risks and harms associated with substance use, we used codes for reasons for and reasons against change that included statements referring to other people in addition to self. Analyses were completed with and without such codes without appreciable differences in results.

² Despite lower coding reliability, scores for commitment language were included in initial predictive analyses. Study results did not differ with subsequent exclusion of commitment language from analyses.

Amrhein, P. C., Miller, W. R., Yahne, C. E., Palmer, M., & Fulcher, L. (2003). Client commitment language during motivational interviewing predicts drug use outcomes. Journal of Consulting and Clinical Psychology, 71(5), 862–878.

Baer, J. S., Garrett, S. B., Beadnell, B., Wells, E. A., & Peterson, P. (2007). Brief motivational intervention with homeless adolescents: Evaluating effects on substance use and service utilization. Psychology of Addictive Behaviors, 21(4), 582–586.

Baer, J. S., & Peterson, P. L. (2002). Motivational interviewing with adolescents and young adults. In W. R.Miller & S.Rollnick (Eds.), Applications of motivational interviewing (2nd ed.). New York: Guilford Press.

Baer, J. S., Peterson, P. L., & Wells, E. A. (2004). Rationale and design of a brief substance use intervention for homeless adolescents. Addiction Research & Theory, 12(4), 317–334.

Cicchetti, D. V. (1994). Guidelines, criteria, and rules of thumb for evaluating normed and standardized assessment instruments in psychology. Psychological Assessment, 6(4), 284–290.

Cohen, J., Cohen, P., West, S. G., & Aiken, L. S. (2003). Applied multiple regression/correlation analyses for the behavioral sciences (3rd ed.). Mahwah, NJ: Erlbaum.

Fishbein, M., & Ajzen, I. (1975). Belief, attitude, intention, and behavior. Reading, MA: Addison-Wesley.

Fishbein, M., Hennessy, M., Yzer, M., & Douglas, J. (2003). Can we explain why some people do and some people do not act on their intentions?Psychology, Health & Medicine, 8(1), 3–18.

Hettema, J., Steele, J., & Miller, W. R. (2005). Motivational interviewing. Annual Review of Clinical Psychology, 1(1), 91–111.

Karver, M. S., Handelsman, J. B., Fields, S., & Bickman, L. (2006). Meta-analysis of therapeutic relationship variables in youth and family therapies: The evidence for different relationship variables in the child and adolescent outcome literature. Clinical Psychology Review, 26(1), 50–65.

Miller, W. R. (2000). Motivational Interviewing Skill Code (MISC): Coder's manual. Retrieved from http://casaa.unm.edu/download/misc1.pdf

Miller, W. R., Benefield, R. G., & Tonigan, J. S. (1993). Enhancing motivation for change in problem drinking: A controlled comparison of two therapist styles. Journal of Consulting and Clinical Psychology, 61(3), 455–461.

Miller, W. R., & Rollnick, S. (2002). Motivational interviewing: Preparing people to change addictive behavior (2nd ed.). New York: Guilford Press.

Miller, W. R., Sovereign, R. G., & Krege, B. (1988). Motivational interviewing with problem drinkers: II. The Drinker's Check-up as a preventive intervention. Behavioural Psychotherapy, 16(4), 251–268.

Monti, P. M., Colby, S. M., & O'Leary, T. A. (2001). Adolescents, alcohol, and substance abuse: Reaching teens through brief interventions. New York: Guilford Press.

Moyers, T. B., Martin, T., Catley, D., Harris, K. J., & Ahluwalia, J. S. (2003). Assessing the integrity of motivational interviewing interventions: Reliability of the motivational interviewing skills code. Behavioural and Cognitive Psychotherapy, 31(2), 177–184.

Moyers, T. B., Martin, T., Christopher, P. J., Houck, J. M., Tonigan, J. S., & Amrhein, P. C. (2007). Client language as a mediator of motivational interviewing efficacy: Where is the evidence?Alcoholism: Clinical and Experimental Research, 31(Suppl. 3), 40S–47S.

Nippold, M. A. (2000). Language development during the adolescent years: Aspects of pragmatics, syntax, and semantics. Topics in Language Disorders, 20(2), 15–28.

O'Kearney, R., & Dadds, M. R. (2005). Language for emotions in adolescents with externalizing and internalizing disorders. Development and Psychopathology, 17(2), 529–548.

Rollnick, S., & Miller, W. R. (1995). What is motivational interviewing?Behavioural and Cognitive Psychotherapy, 23(4), 325–334.

Shirk, S. R., & Karver, M. S. (2003). Prediction of treatment outcome from relationship variables in child and adolescent therapy: A meta-analytic review. Journal of Consulting and Clinical Psychology, 71(3), 452–464.

Sobell, M. B., & Sobell, L. C. (1993). Treatment for problem drinkers: A public health priority. In J. S.Baer & G. A.Marlatt (Eds.), Addictive behaviors across the life span: Prevention, treatment, and policy issues (pp. 138–157). Newbury Park, CA: Sage Publications.

Webb, J. A., Baer, P. E., Getz, J. G., & McKelvey, R. S. (1996). Do fifth graders' attitudes and intentions toward alcohol use predict seventh-grade use?Journal of the American Academy of Child & Adolescent Psychiatry, 35(12), 1611–1617.

Submitted: October 5, 2007 Revised: April 4, 2008 Accepted: April 23, 2008

Application of Item Response Theory to Tests of Substance-Related Associative Memory

Acknowledgement: This research was supported by two grants from the United States Department of Health & Human Services, National Institutes of Health, National Institute on Drug Abuse (DA024659-04 and DA023368-06). We thank Amy Custer for her work on this project.

In the past two decades, implicit memory and cognition approaches have gained substantial popularity in addiction and health behavior research. Focusing on the role of spontaneously activated cognitions on behavior (Stacy & Wiers, 2010; Wiers & Stacy, 2006), researchers have examined automatic/implicit cognitive processes at different levels of analysis, ranging from attention (e.g., Bradley, Field, Mogg, & De Houwer, 2004; Mogg & Bradley, 2002) to memory (e.g., Krank & Goldstein, 2006; Stacy, 1995, 1997) and attitude (e.g., Chassin, Presson, Sherman, Seo, & Macy, 2010; Huijding, de Jong, Wiers, & Verkooijen, 2005; Houben, Havermans, & Wiers, 2010) by using various indirect tests of implicit cognitive processes related to addictive and health behaviors (see Ames et al., 2007; Stacy, Ames, & Grenard, 2006; Stacy & Wiers, 2010, for review). A recent meta-analysis examining the relationship between substance-related implicit cognition and substance use revealed that the substance-related implicit word-association test (WAT) was the best predictor of substance use with the largest effect size (mean r = .38) among other implicit measures (Rooke, Hine, & Thorsteinsson, 2008). Although studies have reported good reliability of WAT (Ames et al., 2007; Preece, 1978), to the best of our knowledge, no comprehensive psychometric evaluations of WAT have been conducted in research on addiction, cognition, or memory. The current study extends previous research by applying a comprehensive item response theory (IRT) framework to understand and improve the psychometric properties of WAT items and estimation of underlying latent traits of alcohol- and marijuana-related associative memory.

The WAT is one of the most commonly used indirect memory tests for assessing the retrieval of preexisting substance-related associations in memory (Stacy, 1995, 1997). In substance-related WAT, a series of substance-related cue words or phrases are presented one by one visually or auditorily, and participants are asked to generate the first word or short phrase that comes to mind when they think of the cue. It is assumed that an association of a cue–target pair gets strengthened with repetitive encounters with a substance-related cue (e.g., “feeling good”) and target behavior (e.g., marijuana use). Therefore, those who frequently engage in substance use are more likely than those who do not to spontaneously think of substance-use behavior in response to substance-related cues in WAT.

Accumulated evidence has shown that substance- or risky behavior-related implicit associative memory, measured by WAT, has strong predictive power for substance use, including alcohol (Ames & Stacy, 1998; Kelly, Masterman, & Marlatt, 2005; Stacy, 1997), marijuana (Ames et al., 2007; Ames & Stacy, 1998; Stacy, 1997), and cigarette use (Grenard et al., 2008; Kelly, Haynes, & Marlatt, 2008), as well as risky sexual behavior (Ames, Grenard, & Stacy, 2013; Grenard, Ames, & Stacy, 2013; Stacy, Ames, Ullman, Zogg, & Leigh, 2006). Given the successful application of WAT to a wide range of issues in health and cognition, it is important to fully understand the psychometric characteristics and construct validity of the measure. The IRT modeling framework provides one of the most comprehensive strategies available to accomplish these goals and has a number of advantages over the traditional classical test theory (CTT) approach (Reise, Ainsworth, & Haviland, 2005).

IRT consists of a series of statistical models specified to describe the probability of endorsing an item as a function of an underlying latent trait (θ). In the context of the alcohol-related WAT, IRT describes the association between the probability that a participant generates an alcohol-related response to a given WAT item and his or her level of the latent alcohol-related implicit associative memory. The use of IRT in psychometric evaluation has several advantages over classical test theory (CTT). First, IRT allows for detailed investigation of WAT items in relation to the latent alcohol-related associative memory. It provides parameter estimates of item difficulty (b) and item discrimination (a). The b parameter indicates (1) how difficult a given WAT item is and (2) what level of latent memory association is needed so that 50% of participants would endorse an alcohol-related response to a given WAT item. Participants whose trait levels (i.e., alcohol-related associative memory) are higher are likely to generate an alcohol-related response to a WAT item with a higher b parameter value. The a parameter tells how effectively a WAT item differentiates among individuals with different levels of latent implicit alcohol-related associative memory. The item with a higher value of a is a good item because such an item discriminates effectively between individuals of slightly different levels of latent alcohol-related implicit associations in memory.

A critical advantage of IRT over CTT is that IRT is sample-invariant whereas CTT is sample-dependent (Hambleton & Jones, 1993). Under the situation in which an IRT model fits the data, the item parameter estimates (i.e., the a and b parameters) can be interpreted independent of the study sample (item-parameter invariance; Lord, 1980). Similarly, a latent trait can be estimated independent of a set of test items used in a study (person-parameter invariance; Lord, 1980). These sample invariant characteristics are not true in CTT. In CTT, item discrimination (i.e., item–total correlation), item difficulty (i.e., proportion of correct) and scale scores (i.e., the summed score) are completely dependent on a sample. Thus, an estimate of a latent trait score in CTT is largely affected by the characteristics of a study sample (Hambleton & Jones, 1993).

Another advantage of IRT is that reliability can be estimated with great flexibility. In CTT, a reliability estimate (e.g., Kuder–Richardson Formula 20, Cronbach’s coefficient alpha) is a fixed constant for all items. In contrast, reliability in IRT can be estimated at any point in the range of an underlying latent trait. Moreover, reliability estimates can be computed at both the item and test levels, using the item information and test information functions (IIF and TIF), respectively. In our substance-related WAT, we determined the extent to which each WAT item and WAT as a test accurately estimated a specific level of implicit substance-related associative memory.

Last, the IRT framework allows for the investigation of differential item functioning (DIF). The DIF analysis assesses whether or not a test item functions equivalently across subgroups of a study sample while controlling for the overall difference in the latent trait levels. For example, if the a or b parameter of a given WAT item is different between male and female participants with the same level of the latent implicit alcohol-related associative memory, the item is considered to exhibit DIF and could be a threat to the construct validity of the alcohol-related WAT (Kristjansson, Aylesworth, McDowell, & Zumbo, 2005).

The current study evaluated psychometric properties of two forms of a substance-related WAT, marijuana- and alcohol-related, using a unidimensional IRT modeling approach. The data were collected as part of a large-scale longitudinal study of dual-process theory and drug use in adolescents and consisted of 775 ethnically diverse, at-risk high school students in Southern California. The adolescent sample was chosen because of sufficient variability in alcohol, marijuana, and other drug use as well as the importance of this age group for the study of drug use progression.

The aims of the study were to (a) evaluate parameters of substance-related WAT items including item difficulty and item discrimination, (b) examine the precision of WAT at the item and test levels, (c) estimate the latent trait scores (i.e., the level of substance-related implicit associative memory) for each participant, and (d) evaluate criterion validity through the association between WAT scores and substance-use measures. Results of comprehensive psychometric validation of substance-related WAT will be discussed in light of the framework of a unified concept of construct validity (Messick, 1975, 1989, 1995). The comprehensive IRT approach illustrated in this article is applicable to a wide variety of measurement issues in associative memory and other areas of addiction and health behavior research.

Participants

The participants were 775 continuation high school (CHS) students (340 female) in the greater Los Angeles area. Their participation in this study did not require current or past history of substance use. The participants’ ages ranged from 14 to 20, of which 94% were between the ages of 15 and 18. The study sample comprised Hispanic (62.5%), Non-Hispanic White (12.5%), mixed race/ethnicity (18.7%), Black (3.2%), and other race/ethnicity that included Asian, Native American, and “other” (3.1%). They were recruited from classes from 42 CHSs, which were selected from over 100 CHSs in the region. The schools sampled did not provide any drug education programs to their students.

Measures

Word-association test

As described in the introduction, the substance-related WAT is an indirect memory test designed to assess the spontaneous retrieval of preexisting substance-related associations in memory (e.g., Stacy, 1997). The current study used two formats of WAT, an outcome-behavior association task (OBAT) and a compound-cue version of WAT. In OBAT, all cues are phrases that are related to affective outcomes of drug use (e.g., “feeling good”). In the compound-cue WAT, cues consist of either a combination of location and affective outcome phrases (two compound cues; e.g., “my bedroom, feeling good”) or a combination of situation, location, and affective outcome phrases (three compound cues; e.g., “weekend, friend’s house, having fun”). Fillers are cues that are unrelated to substance use (e.g., “doing homework”). Each of the three cue types had six target cues and two filler cues, totaling 18 target and six filler cues. Each trial started by visually presenting a cue phrase in the center of a computer display, and participants were instructed to respond with the first behavior or action that came to mind as quickly as possible. Responses were typed in a text box that appeared right below where the cue was presented. The next trial was generated by participants’ clicking a text button that reads “click here to continue” or after 21 seconds elapsed since the presentation of a cue, whichever came first.

The self-coding procedure (Frigon & Krank, 2009; Krank, Schoenfeld, & Frigon, 2010) was employed to code the WAT responses upon completion of the WAT session. In this procedure, participants were presented with a WAT cue and their typed response on the computer display, along with a list of 12 behavior categories (e.g., alcohol, marijuana, tobacco, exercise, etc.). They were asked to check one or more categories that were related to their responses. A checked response was coded 1 and an unchecked response was coded 0, and these scores were summed to yield a total WAT score for each category. In the current study, the scores for alcohol- and marijuana-related responses were examined separately.

Drug use: Marijuana and alcohol

Frequency of drug use was measured by a self-report drug-use questionnaire (Stacy et al., 1990; Stacy, 1997) that asked participants to indicate how many times they had used each drug in the past year and the past 30 days. The questionnaire was an 11-point rating scale, with frequency response options ranging from 1 (None) to 11 (91 + times). The reliability and validity of these self-reported drug-use measures were demonstrated elsewhere (e.g., Stacy et al., 1990).

Other variables

Participants’ demographics (age, gender, and language use), scores on the Rutgers Alcohol Problem Index (RAPI; White & Labouvie, 1989), and frequencies of simultaneous polydrug use (Collins, Ellickson, & Bell, 1998) were also assessed. These measures were used as predictors in missing data analyses reported below (see Data Analysis Plan for more details).

Procedure

We contacted each continuation high school (CHS) to arrange recruitment and obtained both written assent from eligible students and consent from their parents. Assent and consent forms explained that the purpose of the study was to investigate teenagers’ health behaviors, requiring their participation in three assessments over the course of two years to complete the study. Computer-based assessments were administered during regular school hours in groups of up to 20 (M = 11.67 participants per session) in a classroom that was provided by each CHS. Data collectors set up a mobile computer laboratory in each classroom that included 20 laptop computers supplied by the research project. Upon arrival to the laboratory, participants were randomly assigned to a computer. After the instructions were given, the assessments began by participants’ pressing any number key on the keyboard. The rest of the assessments were self-directed by the computer program. A session lasted an average of 60 minutes. Participants received a $10 movie ticket in exchange for their participation during Wave 1; the data reported in this article.

Data analysis plan

Evaluation of the psychometric properties of marijuana- and alcohol-related WAT items consisted of three steps. The analyses were conducted separately for each type of WAT.

IRT assumption checking

We tested two assumptions of the IRT model, unidimensionality and local independence, using both categorical confirmatory factor analysis (CCFA) and IRT methods. In assessing unidimensionality, we fit a one-factor CCFA model with weighted least-squares with mean and variance adjustment (WLSMV; Muthén, du Toit, & Spisic, 1997), by constraining all WAT items to load onto a single latent factor of implicit alcohol-related (or marijuana-related) associative memory. The model fit was evaluated according to the guidelines of Hu and Bentler (1999). A unidimensional two-parameter logistic (2PL) model was also fitted to the data to evaluate unidimensionality. We evaluated overall model fit by examining Maydeu-Olivares-Joe’s M2 (Maydeu-Olivares & Joe, 2006), a limited-information overall fit statistic, as well as the item-level model fit by assessing the S-χ², an item misfit index (Orlando & Thissen, 2000).

The local independence (LI) assumption was evaluated by checking modification indices (MI) of residual covariances in the one-factor CCFA model and the local dependence (LD) statistics (Chen & Thissen, 1997) in the 2PL model. Potential local dependence (LD) is suspected when an excess correlation between a pair of items is observed after controlling for a single latent construct (Thissen & Steinberg, 2009). This suggests a violation of LI, implying to some investigators that the two items ask the exact same questions twice (Varni et al., 2010).

Differential item functioning

DIF tests were conducted to test for item invariance across gender and age groups. Two types of DIF were examined: Uniform DIF implies that the item exhibits a difference in the b parameter between two groups. Nonuniform DIF reflects a difference in the a parameter between two groups. Note that a group difference in DIF is examined while controlling for the overall group difference in the levels of the latent trait. The current study used a one-step Wald test (Cai, Thissen, & du Toit, 2011; Woods, Cai, & Wang, 2013), in which designated anchor items were used to link the latent trait metric for two groups. Anchor items were those items designated not to vary across groups. We identified the anchored items from a two-step Wald test (see Langer, 2008, for more detail) before conducting the one-step Wald test. In the one-step Wald test, a model fit was conducted in the following one-step manner: The mean and standard deviation (SD) of the reference group were fixed to 0 and 1, respectively, and the mean and SD of the focal group and the item parameters (the a and b parameters) were estimated at the same time. The item parameters for the designated anchor items were constrained to be equal between the two groups, whereas those for the candidate items were free to vary between the two groups. The software we used, flexMIRT (Cai, 2012), produces results of the Wald χ² test for the comparisons of the candidate item(s) between the two groups. In comparisons between male and female participants, we used male participants as the reference group. In comparisons between younger (14–16 years old) and older participants (17 years old and above), the reference group was the young group.

IRT: Item-parameter estimation

We evaluated item parameter estimates of any alternative sets of WAT items suggested by the preceding analysis. Both 1PL and 2PL models were fitted to examine whether the a parameter should be fixed or varied across the WAT items. Further investigated was the amount of information each WAT item and the total WAT scale provided with respect to the latent trait. The item containing more information at a given level of the latent trait is considered more reliable. Latent trait scores for alcohol- and marijuana-related implicit associative memory were estimated separately as a function of various WAT-item scores, using expected a posteriori (EAP) estimation.

Criterion-related validity

Criterion-related validity coefficients of substance-related WATs were calculated by separately correlating marijuana- and alcohol-related WAT scores with respective drug-use frequencies from the past 30 days and 1 year. A nonparametric bootstrap method (Efron, 1979, 1987; Efron & Tibshirani, 1985) was used to estimate the Pearson correlation coefficients (termed r*) and their confidence intervals, as the assumption of bivariate normality was violated. We used a bias-corrected and accelerated (BCa) procedure (Efron, 1987) to construct confidence intervals for r* between the following pairs of variables: alcohol-WAT scores and alcohol use from the past 30 days; alcohol-WAT scores and alcohol use from the past year; marijuana-WAT scores and marijuana use from the past 30 days; and marijuana-WAT scores and marijuana use from the past year.

Missing Data

The missing data rates on the WAT ranged from 2% to 15% across 18 items, which was not unexpected with open-ended item formats. In the IRT analyses, list-wise deletion (LWD) of missing data was implemented. The use of LWD in IRT analyses is supported by several IRT simulation studies that have demonstrated acceptable-to-good parameter estimates of item discrimination and difficulty (Finch, 2008), no bias of uniform DIF detection with missing at random (MAR) data (Robitzsch & Rupp, 2009), and very close results to a complete data set (i.e., a data set with no missing data) in terms of power, Type I error rate, and effect sizes in the detection of nonuniform DIF (Finch, 2011).

In the criterion-related validity analysis that was conducted with psychometrically validated WAT items, multiple imputation (MI; Rubin, 1987) was used for missing data to obtain unbiased estimates of parameters. Multivariate imputation by chained equations (van Buuren, Boshuizen, & Knook, 1999; van Buuren & Oudshoorn, 2000) was used as the specific form of multiple imputation, applying the mice package (van Buuren & Groothuis-Oudshoorn, 2011) in the R statistical environment (R Development Core Team, 2012). This technique has recently gained popularity (Azur, Stuart, Frangakis, & Leaf, 2011) due to its ability to model each variable with missing data, regardless of its distribution (see van Buuren & Groothuis-Oudshoorn, 2011, for detailed procedures).

Participants’ demographic variables and their alcohol and marijuana use are summarized in Table 1. To determine whether or not the school-cluster variables should be taken into account in subsequent analyses, we computed the design effect and intraclass correlation for alcohol and marijuana use among the average of 42 CHSs. A design effect of 2.0 was used as a cut-off (see Muthén & Satorra, 1995). The design effect (intraclass correlation in parentheses) for alcohol and marijuana use was 1.8 (.016) and 1.5 (.012), respectively. Thus, the school-cluster variable was not included in our analyses.

Demographic Variables and Substance Use

Alcohol-Related Word Association

IRT assumption checking

Both CCFA and 2PL models showed a good fit to the data, indicating unidimensionality of the 18 alcohol-related WAT items. Results from CCFA, conducted using Mplus, Version 6.11 (Muthén & Muthén, 2011), revealed fit indices as follows: comparative fit index (CFI) = .944, Tucker-Lewis Index (TLI) = .937, and root mean square error of approximation (RMSEA) = .045, with a 90% confidence interval (CI) of .036 to .054. All of the 18 factor loadings were significant (p < .01), ranging from .46 to .75. A 2PL model was fitted using flexMIRT, Version 1.0.4.3 (Cai, 2012) and indicated a good model fit (RMSEA = .04). Regarding the LI assumption, there were three potential item pairs with LD, implied by relatively large values of modification indices (MI) for residual covariances: (1) “friend’s house, feeling a rush” and “weekend, friend’s house, feeling a rush,” (2) “friend’s house, feeling a rush” and “feeling a rush,” and (3) “my bedroom, feeling good” and “my bedroom, feeling relaxed.” In the IRT analysis, no indication of LD item pairs (LD χ² > 10) was obtained. Only one item (“weekend, party, feeling high”) showed a poor item fit (p < .0001). After examining the item contents, we set aside “friend’s house, feeling a rush” and “weekend, party, feeling high,” from a subsequent analysis. The model fit was slightly improved after removing these two items, CFI = .961, TLI = .955, and RMSEA = .037 (90% CI = .027 − .047).

DIF

The DIF test detected only one item exhibiting DIF across gender. The item “weekend, friend’s house, feeling a rush” discriminated more effectively for male (a = 2.51) than female participants (a = 1.27; p < .02). Thus, the item was excluded from the subsequent analysis. We also dropped “feeling high” because the discrimination parameter for the male group was substantially low (a = .78). With regard to age groups, the item “feeling more relaxed” was the only item with a significant uniform DIF (p < .02), indicating that this item was easier for older (b = 1.62) than younger participants (b = 2.31). However, we expected that some items might be more difficult at younger ages while still being potentially applicable to later changes with increasing age. Thus, the item remained in the analysis.

IRT: Item parameter estimation

Our revised alcohol-related WAT, reduced to 14 items (α = .80), was fitted with both 1PL and 2PL models. Both models indicated a good fit (RMSEA = .04 and .03 for 1PL and 2PL, respectively), with no evidence of a violation of LI. A likelihood ratio test revealed a significant improvement in fit by the 2PL, relative to the 1PL, G² (17) = 42.27, p < .001. These results indicated that alcohol-WAT data were reproduced by the model better when the a parameters were estimated freely (2PL), rather than being constrained to be equal (1PL). Table 2 presents the estimated parameters for both models. The common a parameter in 1PL was 2.02. In 2PL, the a parameters ranged from 1.62 to 2.46, indicating that all 14 alcohol WAT items effectively differentiated the participants across different levels of the latent trait. The b parameters in both models were very similar for each item. For most of the WAT items, moderate to strong levels of implicit alcohol-related memory associations were needed to endorse alcohol-related responses. All of these parameter estimates are graphically represented in the item-characteristic curves (ICC; Figure 1).

Item Parameter Estimates for 1PL and 2PL in the Alcohol- and Marijuana-Related WAT

Figure 1. Item characteristic curves for the revised alcohol (solid) and marijuana (dashed) WAT items. The x-axes show the level of theta (latent implicit alcohol- or marijuana-related associative memory), with 0 representing the average level of theta. The y-axes show the proportion of alcohol- or marijuana-related responses generated to each WAT cue.

All ICCs show that the probability of endorsing alcohol-related responses was low for those participants whose latent trait levels were below 1.0. The slopes of most WAT items were steepest throughout the range of the latent levels from about 1.0 to 2.0. These items also provided most information about the latent trait (i.e., most reliable) in this range of the latent level (see Figure 2). The amount of information provided by each WAT item was summed to create the test information curve (TIC; see Figure 3), which demonstrates that the alcohol-related WAT is most reliable at moderate-to-high levels of the latent trait. Estimated latent trait scores (see Table 3) revealed that those who endorsed one alcohol-related response were estimated to possess an average level of latent alcohol-related associative memory. As participants endorsed more alcohol-related responses, their latent score increased.

Figure 2. Item information functions for the revised alcohol (solid) and marijuana (dashed) WAT items. The x-axes show the level of theta (latent implicit alcohol- or marijuana-related associative memory), with 0 representing the average level of theta. The y-axes show the amount of information (I), an index of how accurately each item contributes to estimate the latent trait at a given level of theta.

Figure 3. Test information functions (TIFs) for the revised alcohol (solid) and marijuana (dashed) WATs. The x-axis shows the level of theta (latent implicit alcohol- or marijuana-related associative memory), with 0 representing the average level of theta. The y-axis shows the amount of information (I), an index of how accurately each form of WAT estimates the latent trait at a given level of theta. I = 5, 10, and 20 is equivalent to a reliability estimate of .80, 90, and .95, respectively.

Alcohol- and Marijuana-Related WAT Scores, Latent Trait Scores (EAP) and Their Standard Deviations

Marijuana-Related Word Association

IRT assumption checking

CCFA showed a good fit of the marijuana-model, CFI = .978, TFI = .975, RMSEA = .043 (90% CI = .034–.052). IRT analysis revealed an adequate model fit (RMSEA = .05) with no indication of a poor item fit. Thus, the marijuana-related WAT was determined unidimensional. With regard to LI, both CCFA and 2PL detected only one potential LD item pair, “feeling high” and “my bedroom, feeling high.” After reviewing the item contents, we removed the latter item from the analysis.

DIF

For gender, two items were detected as having uniform DIF: “feeling a rush” (p = .02) and “forgetting problems” (p = .03). The endorsement of the marijuana-related response “Feeling a rush” was easier for females (b = 1.59) than for males (b = 2.03) after matching the two groups on the latent trait. Conversely, “forgetting problems” was easier for males to endorse (b = 1.18) than for females (b = 1.44). These two items were removed from the subsequent analyses. As for age groups, no items exhibited a significant DIF.

IRT: Item parameter estimation

The revised marijuana-related WAT had a total of 15 items (α = .87). Both 1PL and 2PL models fit the data adequately (RMSEA = .05) with no sign of LD item pairs. A likelihood ratio test showed that 2PL had a significantly better fit than 1PL, G² (17) = 77.88, p < .001. Estimated item parameters by both models and ICCs (2PL only) are presented in Table 2 and Figure 1, respectively. In 2PL, the a parameter varied from 1.37 to 3.62 and the b parameters ranged from −.48 to 2.05. As shown in Figure 2, IIFs show that “feeling high” was the only item most reliable at the below-average level of the latent trait. Still, TIF illustrates that the marijuana WAT was most reliable around the moderate-to-high levels of the latent trait continuum (see Figure 3). Estimated latent trait scores showed that a total marijuana WAT score of 3 corresponded to the average level of the latent marijuana-related associative memory. A monotonically increasing relationship was observed between the total WAT scores and the latent trait scores (see Table 3).

Criterion-Related Validity

Table 4 shows the correlations between substance-related WAT scores and drug-use frequencies. In both alcohol- and marijuana-related WATs, the participants who endorsed more substance-related responses tended to report higher frequencies of substance uses both in past year use (r* = .44 [BCa CI = .37–.51] and .56 [BCa CI = .50–.61], for alcohol and marijuana, respectively) and past 30-days use (r* = .38 [BCa CI = .30–.47] and .48 [BCa CI = .42–.54], for alcohol and marijuana, respectively).

Bivariate Correlations (BCa 95% CI) Between the Substance-Related WAT Scores and Frequencies of Past Drug Use

The present study was the first to apply a comprehensive psychometric framework using IRT approaches to evaluate psychometric properties of alcohol- and marijuana-related WATs in a sample of ethnically diverse, at-risk adolescents. Our results have demonstrated that both forms of WAT have good psychometric properties when subjected to a comprehensive latent variable and IRT analyses. The discussion below focuses on key findings regarding item and scale properties as well as evidence of construct validity (Messick, 1989, 1995).

Alcohol- and Marijuana-Related WAT: Scale Properties

The original 18 WAT items were reduced to 14 and 15 items in alcohol- and marijuana-related WAT, respectively. Items were removed because they exhibited poor item fit (two items each in both WATs), LD issues (one item each in both WATs), or gender bias (one item in the alcohol WAT and two items in the marijuana WAT). Excluding these items improved the revised versions of the substance-related WAT. As expected, both forms of WAT were shown to be unidimensional and most reliable with individuals with moderate-to-high levels of latent alcohol- or marijuana-related associative memory (see Figure 3). A monotonically increasing relationship between the total WAT scores and estimated latent trait scores was observed in both WATs (see Table 3). These results confirmed that the substance-related WATs measure a single construct of substance-related associative memory, as it purports to do. Furthermore, the total alcohol- and marijuana-WAT scores were positively correlated with frequencies of respective past substance-use behaviors, providing strong evidence of criterion-related validity. This finding is in agreement with that of Krank et al. (2010), who reported that self-coded WAT scores were positively associated with past 30-days alcohol use among college students and added to evidence supporting the use of self-coded scoring procedures (Frigon & Krank, 2009; Krank et al., 2010).

Alcohol- and Marijuana-Related WAT: Item Properties

Item discrimination for all items in both WATs showed high discrimination parameters (a > 1.35). Among the items, some of the compound cues exhibited very high discrimination parameter values, especially in the marijuana-related WAT. Those compound cues included “friend’s house, having fun” (a = 3.62), “friend’s house, hanging out, feeling good” (a = 3.14), and “Friday night, friend’s house, having fun” (a = 3.20). A possible explanation for this is that when a positive affective outcome cue was combined with a peer cue to create a compound cue, its item discrimination was further improved. This explanation is consistent with some theories of adolescent substance use that focus on peer influence as a pivotal risk factor (e.g., Hawkins, Catalano, & Miller, 1992; Petraitis, Flay, & Miller, 1995).

With respect to item difficulty, most of the WAT items were most reliable at the moderate-to-high levels of the latent construct. However, a slightly different pattern of results was observed across the two WATs. In the alcohol WAT, all but one item (“Friday night, friend’s house, having fun”) had item-difficulty parameter estimates greater than 1.0. This indicates that the probability of endorsing alcohol-related responses to the items was lower than .5 for the participants with below moderate levels of latent alcohol-related associative memory. In contrast, the marijuana-related WAT contained a mix of items with moderate-to-high difficulty parameters and items with lower difficulty parameters. This led the marijuana WAT to cover a wider range of the latent-trait continuum than the alcohol WAT. For example, even among those participants with a lower level of latent marijuana-associative memory, half of them endorsed marijuana-related responses to the cues, “feeling high” (b = −.48) and “weekend, party, feeling high” (b = .20). On the other hand, these two cues were not good ones for alcohol. Both items showed a poor model fit and hence were excluded from the revised alcohol-WAT. Further, the only item with the phrase “feeling high” in the revised alcohol-WAT had a high difficulty parameter estimate (“my bedroom, feeling high,” b = 1.97). Hence, we consider “feeling high” as a cue strongly associated with marijuana, particularly at a lower range of the latent trait. This suggests that inclusion of behavior-specific cues may further improve the psychometric properties of substance-related WAT.

Alcohol and Marijuana-Related WAT: Unified Concept of Validity

Traditionally, construct validity has been examined by use of multitrait-multimethod matrix (MTMM matrix, Campbell & Fiske, 1959) or confirmatory factor analysis (CFA, Jöreskog, 1969; Kenny & Kashy, 1992; Stacy, Widaman, Hays, & DiMatteo, 1985) procedures to gather evidence of convergent and discriminant validity. In contrast, Messick (1989, 1995) suggested six aspects of construct validity, arguing that construct validity of a measurement instrument should be justified by use of the available evidence for a wide variety of aspects of construct validity, including content, substantive, structural, generalizability, external, and consequential aspects. The current study showed that the substance-related WATs exhibited evidence of each of these aspects of construct validity. For example, the content aspect is evidenced by the fact that all substance-related WAT items were selected from, or created based on, past studies that reported the utility of WAT as a measure of substance-related implicit associative memory (e.g., Ames et al., 2007). A unidimensional structure of both forms of WATs supports the structural aspect of construct validity, indicating that a single construct of alcohol- or marijuana-related implicit associative memory is evaluated in the WAT. Regarding the substantive aspect, which requires empirical evidence of response consistencies from data, both forms of WAT revealed good internal consistency across a range of the latent trait. For example, the amount of information (I) exceeded 5.0, which is equivalent to a reliability estimate of .80, at the underlying latent trait levels between 0 and 2.0 (see Figure 3). In terms of the generalizability aspect of construct validity, the DIF tests demonstrated that all items in the revised version of the alcohol- and marijuana-related WATs were invariant across gender and age groups. Finally, although we were not able to investigate any evidence of convergent and discriminant validity in the current study, the obtained evidence of criterion-related validity for both forms of WAT justifies the external aspect of construct validity. As reported above, a significant correlation was found between substance-related WAT scores and frequencies of substance use, both in the past 30 days and the past year. Overall, the current study revealed multiple lines of evidence for the construct validity of the alcohol- and marijuana-related WAT, in accord with the unified concept of construct validity (Messick, 1989, 1995).

Limitations

Several caveats in the present study need to be addressed. First, item invariance was examined only across gender and age groups due to the limited number of samples representing different subgroups (e.g., ethnicity). Thus, the current WAT items might have shown DIF across other subgroups. Future investigations that explore DIF of substance-related WAT items could be conducted across ethnicity and other defining characteristics. Second, because the data were cross-sectional, the direction of the possible causal relationship between WAT scores and past drug use was not inferred. Last, drug-use behavior was measured via a self-report questionnaire, thus responses are sensitive to demand characteristics and/or social desirability bias. However, under circumstances in which adolescents were assured that responses would be confidential, adolescent self-reports have been shown to be accurate (Dent, Sussman, & Stacy, 1997; Donohue, Hill, Azrin, Cross, & Strada, 2007).

Despite these limitations, the present study revealed sound psychometric properties of the alcohol- and marijuana-related WAT. Both forms of WATs were most reliable at moderate-to-high levels of the underlying implicit alcohol- or marijuana-related associative memory. Knowledge of the level of reliability at different levels of the latent trait is one of the several fundamental advantages of IRT over traditional psychometric evaluation (e.g., CTT), in addition to advantages of sample invariance, flexibility, and rigor in evaluating differential item functioning. The IRT and construct validation procedures shown here are useful for a wide range of research topics in addiction as well as basic cognitive research on WAT. Although the procedures can be applied to any presumed measures of an underlying trait, it may be surprising that these highly quantitative procedures can be effectively applied to responses that are self-generated and open-ended—the responses are essentially qualitative in origin. When such responses are amenable to numeric coding, they can be usefully integrated into formal and comprehensive tests of psychometrics and construct validity as revealed here.

¹ In several recent studies by other investigators on continuation high schools (CHSs) in the greater Los Angeles area (e.g., Barnett et al., 2013; Sussman, Sun, Rohrbach, & Spruijt-Metz, 2012), sample characteristics (including the male-to-female ratio, the mean age, racial/ethnic profile, and past alcohol and marijuana use) were very similar to those in the current study. Although demographic information was not available on all CHSs in the region, the general consistency across diverse studies in the region suggests that the present sample is at least similar to other samples previously drawn from the population.

² Although it may appear that the alcohol- and marijuana-related WATs should be analyzed by a multidimensional IRT approach, we used a unidimensional approach, as we used the same set of items for both substances. The use of overlapping items was inevitable to take into account individual differences in substance-related associative memory, as has been evidenced in previous studies (see Stacy, Galaif, Sussman, & Dent, 2006; Sussman, Stacy, Ames, & Freedman, 1998). A compensatory multidimensional model was also not relevant because a WAT response related to one substance should not be compensated by one’s level on the construct of the second substance. Thus, the two forms of WATs were analyzed separately.

Ames, S. L., Grenard, J. L., & Stacy, A. W. (2013). Dual process interaction model of HIV-risk behaviors among drug offenders. AIDS and Behavior, 17, 914–925. doi:10.1007/s10461-012-0140-2

Ames, S. L., Grenard, J. L., Thush, C., Sussman, S., Wiers, R. W., & Stacy, A. W. (2007). Comparison of indirect assessments of association as predictors of marijuana use among at-risk adolescents. Experimental and Clinical Psychopharmacology, 15, 204–218. doi:10.1037/1064-1297.15.2.218

Ames, S. L., & Stacy, A. W. (1998). Implicit cognition in the prediction of substance use among drug offenders. Psychology of Addictive Behaviors, 12, 272–281. doi:10.1037/0893-164X.12.4.272

Azur, M. J., Stuart, E. A., Frangakis, C., & Leaf, P. J. (2011). Multiple imputation by chained equations: What is it and how does it work?International Journal of Methods in Psychiatric Research, 20, 40–49. doi:10.1002/mpr.329

Barnett, E., Spruijt-Metz, D., Unger, J. B., Rohrbach, L. A., Sun, P., & Sussman, S. (2013). Bidirectional associations between future time perspective and substance use among continuation high-school students. Substance Use & Misuse, 48, 574–580. doi:10.3109/10826084.2013.787092

Bradley, B., Field, M., Mogg, K., & De Houwer, J. (2004). Attentional and evaluative biases for smoking cues in nicotine dependence: Component processes of biases in visual orienting. Behavioural Pharmacology, 15, 29–36. doi:10.1097/00008877-200402000-00004

Cai, L. (2012). flexMIRT (TM) Version 1.88: A numerical engine for multilevel item factor analysis and test scoring [Computer software]. Seattle, WA: Vector Psychometric Group.

Cai, L., Thissen, D., & du Toit, S. H. C. (2011). IRTPRO: Flexible, multidimensional, multiple categorical IRT modeling [Computer software]. Lincolnwood, IL: Scientific Software International.

Campbell, D. T., & Fiske, D. W. (1959). Convergent and discriminant validation by the multitrait-multimethod matrix. Psychological Bulletin, 56, 81–105. doi:10.1037/h0046016

Chassin, L., Presson, C. C., Sherman, S. J., Seo, D.-C., & Macy, J. T. (2010). Implicit and explicit attitudes predict smoking cessation: Moderating effects of experienced failure to control smoking and plans to quit. Psychology of Addictive Behaviors, 24, 670–679. doi:10.1037/a0021722

Chen, W. H., & Thissen, D. (1997). Local dependence indexes for item pairs using item response theory. Journal of Educational and Behavioral Statistics, 22, 265–289.

Collins, R. L., Ellickson, P. L., & Bell, R. M. (1998). Simultaneous polydrug use among teens: Prevalence and predictorsJournal of Substance Abuse, 10, 233–253. doi:10.1016/S0899-3289(99)00007-3

Dent, C. W., Sussman, S. Y., & Stacy, A. W. (1997). The impact of a written parental consent policy on estimates from a school-based drug use survey. Evaluation Review, 21, 698–712. doi:10.1177/0193841X9702100604

Donohue, B., Hill, H. H., Azrin, N. H., Cross, C., & Strada, M. J. (2007). Psychometric support for contemporaneous and retrospective youth and parent reports of adolescent marijuana use frequency in an adolescent outpatient treatment population. Addictive Behaviors, 32, 1787–1797. doi:10.1016/j.addbeh.2006.12.005

Efron, B. (1979). Bootstrap methods: Another look at the jackknife. The Annals of Statistics, 7, 1–26. doi:10.1214/aos/1176344552

Efron, B. (1987). Better bootstrap confidence intervals. Journal of the American Statistical Association, 82, 171–185. doi:10.1080/01621459.1987.10478410

Efron, B., & Tibshirani, R. (1985). The bootstrap method for assessing statistical accuracy. Behaviormetrika, 12, 1–35. doi:10.2333/bhmk.12.17_1

Finch, H. (2008). Estimation of item response theory parameters in the presence of missing data. Journal of Educational Measurement, 45, 225–245. doi:10.1111/j.1745-3984.2008.00062.x

Finch, H. (2011). The impact of missing data on the detection of nonuniform differential item functioning. Educational and Psychological Measurement, 71, 663–683. doi:10.1177/0013164410385226

Frigon, A. P., & Krank, M. D. (2009). Self-coded indirect memory associations in a brief school-based intervention for substance use suspensions. Psychology of Addictive Behaviors, 23, 736–742. doi:10.1037/a0017125

Grenard, J. L., Ames, S. L., & Stacy, A. W. (2013). Deliberative and spontaneous cognitive processes associated with HIV risk behavior. Journal of Behavioral Medicine, 36, 95–107. doi:10.1007/s10865-012-9404-6

Grenard, J. L., Ames, S. L., Wiers, R. W., Thush, C., Sussman, S., & Stacy, A. W. (2008). Working memory capacity moderates the predictive effects of drug-related associations on substance use. Psychology of Addictive Behaviors, 22, 426–432. doi:10.1037/0893-164X.22.3.426

Hambleton, R. K., & Jones, R. W. (1993). Comparison of classical test theory and item response theory and their applications to test development. Educational Measurement: Issues and Practice, 12, 38–47. doi:10.1111/j.1745-3992.1993.tb00543.x

Hawkins, J. D., Catalano, R. F., & Miller, J. Y. (1992). Risk and protective factors for alcohol and other drug problems in adolescence and early adulthood: Implications for substance abuse prevention. Psychological Bulletin, 112, 64–105. doi:10.1037/0033-2909.112.1.64

Houben, K., Havermans, R. C., & Wiers, R. W. (2010). Learning to dislike alcohol: Conditioning negative implicit attitudes toward alcohol and its effect on drinking behavior. Psychopharmacology, 211, 79–86. doi:10.1007/s00213-010-1872-1

Hu, L., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling, 6, 1–55. doi:10.1080/10705519909540118

Huijding, J., de Jong, P. J., Wiers, R. W., & Verkooijen, K. (2005). Implicit and explicit attitudes toward smoking in a smoking and a nonsmoking setting. Addictive Behaviors, 30, 949–961. doi:10.1016/j.addbeh.2004.09.014

Jöreskog, K. G. (1969). A general approach to confirmatory maximum likelihood factor analysis. Psychometrika, 34, 183–202. doi:10.1007/BF02289343

Kelly, A. B., Haynes, M. A., & Marlatt, G. A. (2008). The impact of adolescent tobacco-related associative memory on smoking trajectory: An application of negative binomial regression to highly skewed. Addictive Behaviors, 33, 640–650. doi:10.1016/j.addbeh.2007.11.008

Kelly, A. B., Masterman, P. W., & Marlatt, G. A. (2005). Alcohol-related associative strength and drinking behaviours: Concurrent and prospective relationships. Drug and Alcohol Review, 24, 489–498. doi:10.1080/09595230500337675

Kenny, D. A., & Kashy, D. A. (1992). Analysis of the multitrait-multimethod matrix by confirmatory factor analysis. Psychological Bulletin, 112, 165–172. doi:10.1037/0033-2909.112.1.165

Krank, M. D., & Goldstein, A. L. (2006). Adolescent changes in implicit cognitions and prevention of substance abuse. In R. W.Wiers & A. W.Stacy (Eds.), Handbook of implicit cognition and addiction (pp. 439–454). Thousand Oaks, CA: Sage. doi:10.4135/9781412976237.n28

Krank, M. D., Schoenfeld, T., & Frigon, A. P. (2010). Self-coded indirect memory associations and alcohol and marijuana use in college students. Behavior Research Methods, 42, 733–738. doi:10.3758/BRM.42.3.733

Kristjansson, E., Aylesworth, R., McDowell, I., & Zumbo, B. D. (2005). A Comparison of Four Methods for Detecting Differential Item Functioning in Ordered Response Items. Educational and Psychological Measurement, 65, 935–953. doi:10.1177/0013164405275668

Langer, M. (2008). A reexamination of Lord’s Wald test for differential item functioning using item response theory and modern error estimation. Unpublished doctoral dissertation, University of North Carolina at Chapel Hill.

Lord, F. M. (1980). Applications of item response theory to practical testing problem. Hillsdale, NJ: Erlbaum.

Maydeu-Olivares, A., & Joe, H. (2006). Limited information goodness-of-fit testing in multidimensional contingency tables. Psychometrika, 71, 713–732. doi:10.1007/s11336-005-1295-9

Messick, S. (1975). The standard problem: Meaning and values in measurement and evaluation. American Psychologist, 30, 955–966. doi:10.1037/0003-066X.30.10.955

Messick, S. (1989). Validity. In R. L.Linn (Ed.), Educational measurement (pp. 13–103). New York, NY: Macmillan.

Messick, S. (1995). Validity of psychological assessment: Validation of inferences from persons’ responses and performances as scientific inquiry into score meaning. American Psychologist, 50, 741–749. doi:10.1037/0003-066X.50.9.741

Mogg, K., & Bradley, B. P. (2002). Selective processing of smoking-related cues in smokers: Manipulation of deprivation level and comparison of three measures of processing bias. Journal of Psychopharmacology, 16, 385–392. doi:10.1177/026988110201600416

Muthén, B., du Toit, S. H. C., & Spisic, D. (1997). Robust inference using weighted least squares and quadratic estimating equations in latent variable modeling with categorical and continuous outcomes. Unpublished technical report. Retrieved from http://pages.gseis.ucla.edu/faculty/Muthén/articles/Article_075.pdf

Muthén, B., & Satorra, A. (1995). Complex sample data in structural equation modeling. In P. V.Marsden (Ed.), Sociological methodology (pp. 267–316). Oxford, England: Blackwell. doi:10.2307/271070

Muthén, L. K., & Muthén, B. O. (1998–2011). Mplus user’s guide (6th ed.). Los Angeles, CA: Authors.

Orlando, M., & Thissen, D. (2000). Likelihood-based item-fit indices for dichotomous item response theory models. Applied Psychological Measurement, 24, 50–64. doi:10.1177/01466216000241003

Petraitis, J., Flay, B. R., & Miller, T. Q. (1995). Reviewing theories of adolescent substance use: Organizing pieces in the puzzle. Psychological Bulletin, 117, 67–86. doi:10.1037/0033-2909.117.1.67

Preece, P. F. (1978). Three-year stability of certain word-association indices. Psychological Reports, 42, 25–26. doi:10.2466/pr0.1978.42.1.25

R Development Core Team. (2012). R: A language and environment for statistical computing [Computer software] Vienna, Austria: R Foundation for Statistical Computing.

Reise, S. P., Ainsworth, A. T., & Haviland, M. G. (2005). Item response theory: Fundamentals, applications, and promise in psychological research. Current Directions in Psychological Science, 14, 95–101. doi:10.1111/j.0963-7214.2005.00342.x

Robitzsch, A., & Rupp, A. A. (2009). Impact of missing data on the detection of differential item functioning: The case of Mantel-Haenszel and logistic regression analysis. Educational and Psychological Measurement, 69, 18–34. doi:10.1177/0013164408318756

Rooke, S. E., Hine, D. W., & Thorsteinsson, E. B. (2008). Implicit cognition and substance use: A meta-analysis. Addictive Behaviors, 33, 1314–1328. doi:10.1016/j.addbeh.2008.06.009

Rubin, D. B. (1987). Multiple imputation for nonresponse in surveys. New York, NY: Wiley. doi:10.1002/9780470316696

Stacy, A. W. (1995). Memory association and ambiguous cues in models of alcohol and marijuana use. Experimental and Clinical Psychopharmacology, 3, 183–194. doi:10.1037/1064-1297.3.2.183

Stacy, A. W. (1997). Memory activation and expectancy as prospective predictors of alcohol and marijuana use. Journal of Abnormal Psychology, 106, 61–73. doi:10.1037/0021-843X.106.1.61

Stacy, A. W., Ames, S. L., & Grenard, J. L. (2006). Word association tests of associative memory and implicit processes: Theoretical and assessment issues. In R. W.Wiers & A. W.Stacy (Eds.), Handbook of implicit cognition and addiction (pp. 75–90). Thousand Oaks, CA: Sage. doi:10.4135/9781412976237.n6

Stacy, A. W., Ames, S. L., Ullman, J. B., Zogg, J. B., & Leigh, B. C. (2006). Spontaneous cognition and HIV risk behavior. Psychology of Addictive Behaviors, 20, 196–206. doi:10.1037/0893-164X.20.2.196

Stacy, A. W., Flay, B. R., Sussman, S., Brown, K. S., Santi, S., & Best, J. A. (1990). Validity of alternative self-report indices of smoking among adolescents. Psychological Assessment: A Journal of Consulting and Clinical Psychology, 2, 442–446. doi:10.1037/1040-3590.2.4.442

Stacy, A. W., Galaif, E. R., Sussman, S., & Dent, C. W. (1996). Self-generated drug outcomes in high-risk adolescents. Psychology of Addictive Behaviors, 10, 18–27. doi:10.1037/0893-164X.10.1.18

Stacy, A. W., Widaman, K. F., Hays, R., & DiMatteo, M. R. (1985). Validity of self-reports of alcohol and other drug use: A multitrait-multimethod assessment. Journal of Personality and Social Psychology, 49, 219–232. doi:10.1037/0022-3514.49.1.219

Stacy, A. W., & Wiers, R. W. (2010). Implicit cognition and addiction: A tool for explaining paradoxical behavior. Annual Review of Clinical Psychology, 6, 551–575. doi:10.1146/annurev.clinpsy.121208.131444

Sussman, S., Stacy, A. W., Ames, S. L., & Freedman, L. B. (1998). Self-reported high-risk locations of adolescent drug use. Addictive Behaviors, 23, 405–411. doi:10.1016/S0306-4603(97)00069-5

Sussman, S., Sun, P., Rohrbach, L. A., & Spruijt-Metz, D. (2012). One-year outcomes of a drug abuse prevention program for older teens and emerging adults: Evaluating a motivational interviewing booster component. Health Psychology, 31, 476–485.

Thissen, D., & Steinberg, L. (2009). Item response theory. In R.Millsap & A.Maydeu-Olivares (Eds.), The Sage handbook of quantitative methods in psychology (pp. 148–177). London, UK: Sage. doi:10.4135/9780857020994.n7

van Buuren, S., Boshuizen, H. C., & Knook, D. L. (1999). Multiple imputation of missing blood pressure covariates in survival analysis. Statistics in Medicine, 18, 681–694. doi:10.1002/(SICI)1097-0258(19990330)18:6<681::AID-SIM71>3.0.CO;2-R

van Buuren, S., & Groothuis-Oudshoorn, K. (2011). MICE: Multivariate imputation by chained equations in R. Journal of Statistical Software, 45, 1–67.

van Buuren, S., & Oudshoorn, K. (2000). Multivariate imputation by chained equations: MICE V1.0 User’s manual. Leiden, the Netherlands: TNO Prevention and Health. (TNO Report PG/VGZ/00.038).

Varni, J. W., Stucky, B. D., Thissen, D., DeWitt, E. M., Irwin, D. E., Lai, J., . . .DeWalt, D. A. (2010). PROMIS pediatric pain interference scale: An item response theory analysis of the pediatric pain item bank. The Journal of Pain, 11, 1109–1119. doi:10.1016/j.jpain.2010.02.005

White, H. R., & Labouvie, E. W. (1989). Towards the assessment of adolescent problem drinking. Journal of Studies on Alcohol, 50, 30–37.

Wiers, R. W., & Stacy, A. W. (2006). Implicit cognition and addiction. Current Directions in Psychological Science, 15, 292–296. doi:10.1111/j.1467-8721.2006.00455.x

Woods, C. M., Cai, L., & Wang, M. (2013). The Langer-improved Wald test for DIF testing with multiple groups: Evaluation and comparison to two-group IRT. Educational and Psychological Measurement, 73, 532–547. doi:10.1177/0013164412464875

Submitted: May 9, 2013 Revised: December 20, 2013 Accepted: December 27, 2013

Application of the Social Action Theory to Understand Factors Associated With Risky Sexual Behavior Among Individuals in Residential Substance Abuse Treatment

Acknowledgement: This work was supported by NIDA Grant R01 DA19405 and NIDA Grant 1 F31 DA023302–01A1. NIDA had no further role in study design; in the collection, analysis, and interpretation of data; in the writing of the report; or in the decision to submit the paper for publication.

Despite advances in HIV prevention efforts, an estimated 1.1 million individuals are living with HIV/AIDS in the United States (Centers for Disease Control and Prevention, 2008), and approximately 56,300 new infections occurred in 2006 (Hall et al., 2008). Risky sexual behavior (RSB) is currently the most common means of acquiring HIV; approximately 83% of new infections are acquired through sexual transmission (Centers for Disease Control and Prevention, 2008). Individuals who engage in illicit substance use are especially vulnerable to the contraction of HIV through RSBs (e.g., Hoffman, Klein, Eber, & Crosby, 2000). Risk is increased further when considering individuals living in low-income urban areas where HIV risk factors including injection drug use and prostitution occur at higher rates than most other settings (Adimora, & Schoenbach, 2005; Rhodes, Singer, Bourgois, Friedman, & Strathdee, 2005). Given the elevated HIV risk via sexual transmission in urban, substance using populations, there is a clear need to understand the processes underlying RSB in this specific population.

In previous research, a variety of variables have been found to be related to RSB including demographic/background variables such as age, gender, ethnicity, and education (e.g., Avants, Marcotte, Arnold, & Margolin, 2003; Ensminger, Anthony, & McCord, 1997; Johnson, Cunningham-Williams, & Cottler, 2003; Miller & Neaigus, 2002), HIV status (Kalichman, Rompa, & Webster, 2002; Semple, Patterson, & Grant, 2000), past experience of sexual or emotional abuse (Arriola, Louden, Doldren, & Fortenberry, 2005; Senn, Carey, Vanable, Coury-Doniger, & Urban, 2006), disinhibition (Hayaki, Anderson, & Stein, 2006; Lejuez, Simmons, Aklin, Daughters, & Dvir, 2004), negative emotionality and affect (Lehrer, Shrier, Gortmaker, & Buka, 2006; Mustanski, 2007), psychopathology (Johnson et al., 2003; McMahon, Malow, Devieux, Rosenberg, & Jennings, 2008), general self-esteem (Semple, Grant, & Patterson, 2005), partner type (Macaluso, Demand, Artz, & Hook, 2000), and attitudes towards condoms (Somlai et al., 2000). This body of work provides important insight into understanding RSB among urban substance users; yet, there remain open directions for future inquiry. Specifically, there is a need to utilize a theoretical framework to understand the interplay of key variables underlying RSB among urban substance users.

In considering frameworks to understand and intervene on RSB, social–cognitive based models most often have been applied and used to guide interventions. Social–cognitive theory highlights ways in which peoples' beliefs about their personal capabilities, or self-efficacy, influence their attempts to alter ingrained habits (Bandura, 1986). Social–cognitive informed interventions most frequently include the provision of basic HIV information, risk personalization, modeling, skills building (including problem solving and relapse prevention) and to a lesser extent social support enhancement (van Empelen et al., 2003). Although these social–cognitive interventions have been shown to be relatively successful in reducing RSB among drug users (see van Empelen et al., 2003, for a review), there is a clear need for further improvement. Specifically, researchers have identified that the social–cognitive models have failed to account for self-regulatory processes and contextual factors (Bagozzi, 1992; Gollwitzer, 1990; Schwarzer, 1992). Yet, these internal and environmental context factors need to be altered or mobilized to support health risk behavior self-change (Ewart, 2009) and may be of particular importance for urban substance users considering noted difficulties in self-regulation (Palfai, 2006) and high environmental risk (Adimora & Schoenbach, 2005).

Moving in this direction, the social action theory (SAT), an integrative systems model of social-motivational, cognitive, and environmental processes, may provide a novel and potentially productive framework for understanding RSB among urban substance users (Ewart, 1991, in press). SAT was developed as an extension of individual-level psychological theories to address the broad complexities of public health problems. The overarching goal is the detection and manipulation of environmental and self-regulatory skills/deficits that can promote health and/or hinder health behaviors and habits (Johnson, Carrico, Chesney, & Morin, 2008). Initial conceptualizations of SAT focused on behavioral health more broadly defined. From these original roots, SAT has been applied to HIV risk reduction and the prevention of HIV risk behavior (Gore-Felton et al., 2005; Lightfoot, Rotheram-Borus, Milburn, & Swendeman, 2005). For example, previous work has identified the domains of SAT, including contextual and self-regulatory skills/deficits, to be predictive of youth sexual behavior (Mellins, Dolezal, Brackis-Cott, Nicholson, & Meyer-Bahlburg, 2007).

SAT elaborates on existing HIV-related social–cognitive models that emphasize cognitive appraisals and beliefs (e.g., Bandura, 1994; Fisher, Fisher, Williams, & Malloy, 1994) by focusing on social motivational and contextual influences that energize and shape behavior, specifically highlighting the ways in which important self-goals, frequently practiced routines, social-emotional competence, and social power affect substance use and other risky behaviors (Ewart, in press; Lightfoot et al., 2005). Applied to RSB, SAT views the choice of engaging in RSB as influenced by the personal regulatory resources and social power afforded by environmental context (variables that have historical impact on the individual as well as those relevant in the immediate context of the RSB) in combination with the individual's psychopathology and affect, HIV-related attitudes and knowledge, and self-regulatory skills/deficits.

Using SAT as a guiding framework, the current study sought to examine variables associated with event-level RSB among residents of an urban, substance-use treatment facility. In addition to demographic variables and HIV status, predictor variables used for this study were derived from four key domains of SAT: (1) environmental influences (i.e., childhood trauma and characteristics of last sexual intercourse), (2) forms of psychopathology and affect (substance use, depression, anxiety, borderline and antisocial personality disorders, and negative emotionality), (3) HIV-related attitudes and knowledge (condom attitudes and HIV knowledge), and (4) self-regulatory skills/deficits (trait nonplanning impulsivity, delay discounting, and risk-taking propensity). From these more general domains it is important to consider specific variables that capture the relevance of SAT to RSB.

To assess environmental influences on RSB, we utilized a measure of historical experience that would capture physical, emotional, and sexual childhood abuse. We also assessed the characteristics of the last sexual intercourse (using a measure adapted from Tortu, McMahon, Hamid, & Neaigus, 2000) that included variables such as most recent partner type, whether this partner is an injection drug user, HIV positive, and/or having sexual intercourse with other people, as well as whether the participant was drunk or high at the last sexual encounter. These two measures capture historical as well as current environmental influences that may dispose one to engage in RSB by shaping social motives, self-regulatory skills/deficits, and social power in risky situations.

The components of the second domain of SAT, forms of psychopathology and affect, are conceptualized as factors that may interfere with one's ability to self-regulate (i.e., engage in safe sexual practices). To assess psychopathology and affect, we used the Structured Clinical Interview for Diagnostic and Statistical Manual of Mental Disorders (4th ed., DSM–IV; American Psychiatric Association), nonpatient version (SCID–IV–NP; First, Spitzer, Gibbon, & Williams, 2001) to assess for the presence of Axis I and Axis II disorders that may be related to RSB engagement. Specifically, the diagnoses most related to internal-affective states included mood and anxiety disorders, borderline personality disorder (BPD), and antisocial personality disorder (ASPD). In addition to diagnostic status, we also included continuous measures of substance use frequency as well as negative emotionality. Coupling diagnostic information with continuous information related to substance use frequency and affect captures a full picture of the psychopathology/affect domain conceptualized by SAT as being related to RSB.

To assess forms of knowledge and attitudes related to HIV, we utilized the Attitudes Toward Condom Scale (Somlai et al., 2000), which assesses not only negative attitudes related to condom use, but also perceived barriers related to condom nonuse. We also assessed HIV knowledge using an adapted survey from the Center for AIDS Intervention Research (CAIR; see Somlai et al., 2000), which assessed participants' understanding of HIV/AIDS risk, risk reduction steps, and condom usage. The combination of assessing condom attitudes and HIV knowledge encompasses the third domain of SAT related to RSB.

In the assessment of the final domain of SAT, self-regulatory skills/deficits, we used a battery of disinhibition measures. We included trait nonplanning impulsivity (which assesses not planning for the future, lack of self-control, and intolerance of cognitive complexity) given its link to self-regulatory skills/deficits defined by the SAT. We assessed delay discounting using the Delay Discounting Task (DDT; Kirby, Petry, & Bickel, 1999), which captures the degree to which an individual shows preference for either small, immediate rewards or larger, delayed rewards, and fits with the social-emotional competence component of SAT given its relation to factors associated with condom use decision making. To supplement self-report assessment with behavioral assessment of disinhibition, we also utilized the Balloon Analogue Risk Task (BART; Lejuez et al., 2002) to assess risk-taking propensity, a variable previously demonstrated to be associated with RSB in this type of sample (Lejuez et al., 2004). SAT considers RSB to result from self-regulatory skills/deficits as they relate to motivation and problem solving (Ewart, 1991, in press); this battery of behavioral and self-report measures of disinhibition address the self-regulatory skills/deficits delineated by SAT to be related to RSB.

Regarding our main outcome, RSB is often assessed using global association measures that include questioning one's general frequency of RSB during a given period of time (e.g., 6 months; Ross, Hwang, Zack, Bull, & Williams, 2002; Sanchez, Comerford, Chitwood, Fernandez, & McCoy, 2002; Somlai, Kelly, McAuliffe, Ksobiech, & Hackl, 2003). Although useful in many respects, global assessment of past frequency of RSB is subject to inaccuracy of recall and gross frequency count biases (Chawarski, Pakes, & Schottenfeld, 1998). Further, this method of assessment provides little information in terms of situational or context-specific conditions, resulting in condom use being treated a context-independent outcome (Kiene, Barta, Tennen, & Armeli, 2009; Weinhardt & Carey, 2000).

To address the limitations of global assessment of RSB, the utilization of event-level measurement (Leonard, & Ross, 1997; Temple, & Leigh, 1992) has been introduced into research concerning HIV risk behavior (e.g. Gillmore et al., 2002; Tortu et al., 2000). Event-level measurement typically assesses particular aspects of a recent sexual event, such as type of partner and whether one was under the influence of alcohol. In event-level measurement, participants describe behavior during a particular event rather than report general trends or averages of behavior (LaBrie, Earleywine, Schiffman, Pederson, & Marriot, 2005). Despite the potential limitations of underreporting due to forgetting of events (Schroder, Carey, & Vanable, 2003) and the lack of estimate of overall likelihood of RSB, event-level measurement is thought to avoid overreporting and rounding and allows for an examination of situational variables (Kiene et al., 2009).

Few studies have used this event-level methodology with at-risk samples of substance users and within a more comprehensive theory-driven model of risk. Indeed, event-level studies largely are limited to samples including college students (e.g., Brown, & Vanable, 2007; LaBrie et al., 2005) and nonsubstance dependent adult samples (e.g., Ibañez, Van Oss Marin, Villarreal, & Gomez, 2005; Leonard, & Ross, 1997). Of the studies that have utilized event-level data among substance users (e.g., Leigh, Ames, & Stacy, 2008; Scheidt, & Windle, 1996; Tortu et al., 2000; Watkins, Metzger, Woody, & McLellan, 1993), the predictors have largely focused on the effect of intoxication, sexual history (Scheidt, & Windle, 1996), and drug use severity (Scheidt & Windle, 1996; Watkins et al., 1993) and have rarely been incorporated into a broader model of risk. In sum, the current study examined four components of SAT in relation to event-level CU among a sample of substance users in residential substance abuse treatment.

Participants

Participants included 161 men and women receiving treatment at the Salvation Army Harbor Light residential substance abuse treatment facility located in Northeast Washington, DC. Five individuals were found to meet DSM–IV criteria for psychosis and thus were excluded from further analyses. The final sample (n = 156) ranged in age from 18 to 60 years, with a mean age of 42.85 years (SD = 8.59). Of the sample, 54.6% were crack/cocaine dependent, 15.1% heroin dependent, 10.6% marijuana dependent, and 34.3% alcohol dependent (of note adds up to more than 100% due to the fact that participants could be dependent on more than one drug). There were 25.0% who did not meet any dependence criteria, 44.3% were dependent on one drug, 23.6% were dependent on two drugs, 5.0% were dependent on three drugs, and 2.1% were dependent on four drugs. With regard to racial/ethnic background, 86.5% of the participants were African American, 7.1% were White, 1.9% were Hispanic/Latino, 1.3% were Native American, and 3.2% reported “other.” In terms of highest education level, 20.0% had not completed high school, 40.0% completed high school (or received a GED), and 40.0% completed at least some college, technical, or trade school. The majority of the sample reported current unemployment, and 53.9% reported a household income of less than $10,000 a year. Patients entered the treatment center either voluntary or under a pretrial-release-to-treatment program through the District of Columbia Pretrial Services Agency (53.3%). This program offers drug offenders who are awaiting trial the option to receive substance abuse treatment as a way to ensure appearance in court, provide community safety, and address an underlying cause of recidivism. Patients were contracted to a specific length of stay on entry into the treatment center. For the current sample, contract lengths included 30 days (41.4%), 60 days (29.7%), 90 days (6.3%), or 180 days (22.6%).

Procedures

Assessment sessions were held on Friday afternoons in a private room at the Salvation Army Harbor Light facility. Residents at the treatment center were approached and asked if they would be willing to participate in a study examining sexual behavior among substance users. They were told that the session would last up to 3 hr and that they would be paid $25 in the form of a grocery store gift card on discharge from treatment. Interested participants were given a more detailed explanation of the procedures and asked to provide written informed consent approved by the University of Maryland Institutional Review Board. Given issues of reading comprehension, efforts were made to ensure that participants understood all facets of the consent form and the study itself. Next, a diagnostic interview, the SCID–IV–NP (First et al., 2001), was administered in a separate private room by a trained senior research staff member. If not eligible following the SCID–IV–NP due to the presence of psychotic symptoms, participants were debriefed and paid $10 in grocery gift cards for their time. Following completion of the diagnostic interview, the participants completed a battery of questionnaires (described in detail below).

Although the participants were completing the questionnaires, individuals trained in administering the behavioral task took participants one by one into an adjacent room where they completed the computer task (i.e., BART). The order of completion of the questionnaires was counterbalanced across participants. For entry into the treatment center (independent of the current study), individuals were required to evidence abstinence from all substance use and to have completed a detoxification program if needed. As a result, acute drug effects likely did not affect the participants' scores on the testing battery. Once the participants completed each aspect of the assessment session (i.e., clinical interview, questionnaire packet, and computer task), they were told how much money they earned and signed a receipt. This payment ($25 in grocery store gift card) was deposited into their account at Harbor Light, which they received on discharge from the residential treatment center.

Measures

Event-level assessment of CU (event-level CU)

Participants reported whether a condom was used during their last sexual intercourse. This question, based on work by Tortu et al. (2000), served as the dichotomous dependent variable (coded: 0 = condom used, 1 = condom not used).

Demographic information and HIV status

A short self-report questionnaire was administered to obtain age, gender, race, education level, marital status, and total yearly income. In addition, participants reported their HIV status.

SAT derived independent variables

Domain 1: Environmental influences

Childhood trauma

As a measure of childhood trauma, we used the three abuse subscales (physical, emotional, and sexual) of the short form of the Childhood Trauma Questionnaire (Bernstein, Stein, & Newcomb, 2003). The CTQ–SF assesses childhood maltreatment experiences (i.e., “while you were growing up”) using a 5-point scale ranging from 1 (never true) to 5 (very often true). The CTQ–SF shows convergent and discriminant validity with other trauma measures (Bernstein et al., 1994; Fink, Bernstein, Handelsman, Foote, & Lovejoy, 1995). Internal consistency for the scale was adequate with Cronbach's α = .94.

Characteristics of last sexual intercourse

Based on Tortu et al. (2000), we administered a 10-item questionnaire on one's most recent sexual intercourse. Contextual predictors of CU from this measure included: type of partner that was involved in the most recent sexual intercourse encounter (commercial: defined as money or drugs being exchanged for sex; casual: defined as no committed relationship; or regular: defined as a committed relationship including a boyfriend/girlfriend or spouse), as well as whether this most recent partner was an injection drug user, HIV positive, and/or having sexual intercourse with other people. Further, the respondent was asked to indicate whether at their last sexual encounter they were drunk and/or high, and if they also engaged in oral or anal sex. Finally, the questionnaire asked whether the sexual encounter occurred in the last year.

Domain 2: Psychopathology and affect

Psychopathology

The SCID–IV–NP (First et al., 2001) was used to determine the presence of current Axis I disorders (i.e., clinical disorders) including depression, anxiety (presence of any anxiety disorder including panic, social phobia, generalized anxiety disorder, specific phobia, posttraumatic stress disorder), psychosis (for exclusion criteria), and substance dependence as well as Axis II disorders (i.e., personality disorders) including borderline personality disorder (BPD) and antisocial personality disorder (ASPD). This measure has demonstrated reliability (First et al., 2001). All eligible participants were administered the interviews in a private area by trained research staff.

Substance use frequencies

Frequency of drug (i.e., marijuana, heroin, and crack/cocaine) and alcohol use was assessed with a standard substance use questionnaire modeled after the Alcohol Use Disorders Identification Test (AUDIT; Saunders, Aasland, Babor, de la Fuente, & Grant, 1993). Specifically, participants were asked how often they used each substance in the past year prior to treatment. Response options were: 0 (never), 1 (one time), 2 (monthly or less), 3 (2 to 4 times a month), 4 (2 to 3 times a week), and 5 (4 or more times a week). This measure was used to supplement dependence diagnoses from the SCID–IV–NP with more detailed information on substance use frequency.

Negative emotionality

The Multidimensional Personality Questionnaire Brief Form (MPQ–BF; Patrick, Curtin, Tellegen, 2002) is a 155-item version of the original 300-item MPQ (Tellegen, 1982) developed to assess a variety of personality traits and temperamental dispositions. The higher order factor of Negative Emotionality (NEM) is comprised of the traits of Stress Reactivity, Alienation, and Aggression, and we utilized only this subscale given its previously demonstrated link to RSB (Lehrer et al., 2006). Internal consistency for the scale was good (α = .89).

Domain 3: HIV-related attitudes and knowledge

Condom attitudes

The Attitudes Toward Condoms Scale (Somlai et al., 2000) consists of eight items focusing on negative attitudes regarding CU and beliefs concerning barriers to CU, such as perceptions and connotations of CU, perceived unreliability of condoms in preventing sexually transmitted diseases, and embarrassment when purchasing condoms. The scale allows for level of agreement with each item using a 6-point Likert scale ranging from 1 (strongly disagree) to 6 (strongly agree). A participant's negative attitudes toward condoms scale score reflects the mean of the eight items, with higher values reflecting more negative attitudes toward CU. Internal consistency for the scale was acceptable (α = .66).

HIV knowledge

Adapted from a survey developed by CAIR (see Somlai et al., 2000), participants' understanding of HIV/AIDS risk, risk reduction steps, condom usage, and safer sex practices was assessed using a 22-item true–false scale. Scores on this scale range from 0 to 22, reflecting the number of questions correctly answered. Internal consistency for the scale was acceptable (α = .67).

Domain 4: Self-regulatory skills/deficits

Trait nonplanning impulsivity

Trait impulsivity was assessed using the Barratt Impulsiveness Scale (Version 11, BIS–11; Patton, Stanford, & Barratt, 1995). The BIS–11 is a 30-item, self-report questionnaire that contains three subscales, which have been termed Motor Impulsiveness, Cognitive Impulsiveness, and Nonplanning. We utilized only the Nonplanning subscale, which we conceptualized as being most relevant to condom nonuse given its focus on a lack of planning for the future, specifically targeting lack of self-control and intolerance of cognitive complexity. The questions require participants to rate how often a statement applies to them based on the following scale: 1 (rarely/never), 2 (occasionally), 3 (often), to 4 (always/almost always). The BIS–11 has been normed on a variety of sample populations, including college students, inpatient substance abusers, and prison inmates. Internal consistency for the scale was acceptable with Cronbach's α = .68.

Delay discounting

Developed by Kirby et al. (1999), the DDT provides a measure of the degree to which an individual shows preference for either small, immediate rewards or larger, delayed rewards, which may be stated as the rate at which the subjective value of deferred rewards decreases as a function of the delay until they are received. The DDT was provided in a paper-and-pencil format, consisting of a fixed set of 27 choices between smaller, immediate rewards and larger delayed rewards. For example, on the first trial, participants were asked “Would you prefer $54 today, or $55 in 117 days?” Delays included in this questionnaire ranged from 7 to 186 days. The presentation order of the delays was contrived so as not to correlate choice amounts, ratios, differences, delays, or discount rate implied by indifference to the two rewards. Previous research has shown that individual's discount curves are well described by the hyperbolic discount function (Mazur, 1987) V = A/(1 + kD), where V is the present value of the delayed reward A at delay D, and k is a free parameter that determines the discount rate. As k increases, the person discounts the future more steeply. Therefore, k can be thought of as an impulsiveness parameter, with higher values corresponding to higher levels of impulsiveness.

Risk-taking propensity

The BART (Lejuez et al., 2002) was used as a behavioral measure of risk-taking propensity. In this task, the computer screen displays a small simulated balloon accompanied by a balloon pump. Participants are directed to pump the simulated balloon to earn as much money as possible, taking into consideration that the balloon can pop at any time. When a balloon explodes, all money in the temporary bank is lost and the next uninflated balloon appears on the screen. At any point during each balloon trial, the participant can stop pumping the balloon by clicking a “Collect $$$” button, which transfers all money from a temporary bank to a permanent bank. After each balloon explosion or money collection, the participant's exposure to that balloon will end, and a new balloon will appear until a total of 20 balloons (i.e., trials) are completed. These 20 trials are comprised of different balloon types, all with the same probability of exploding. Participants are not given any detailed information about the probability of an explosion, but are told that at some point each balloon will explode and this explosion can occur as early as the first pump all the way up to the point at which the balloon expands as large as the computer screen. The primary dependent measure on the BART was the adjusted number of pumps across trials. An independent review by Harrison, Young, Butow, Salkeld, and Soloman (2005) of state of the art risk measurement strategies identified the BART has having excellent reliability (test–retest) and validity (including convergent and divergent).

Data Analytic Strategy

Analyses were conducted with the dichotomous variable of CU versus nonuse as the dependent variable. Primary analyses began with descriptive statistics for the entire sample across the dependent variable of event-level CU as well as each of the variables in the four domains of SAT: (1) environmental influences, (2) psychopathology and affect, (3) HIV-related attitudes and knowledge, and (4) self-regulatory skills/deficits. Next, we examined differences between those who used a condom versus those who did not at the last sexual encounter on all study variables using paired samples t tests and chi-square. Each variable found to significantly differ for the CU versus nonuse groups were included in a multivariate logistic regression analysis. All analyses used a two-tailed alpha of .05.

As a preliminary analysis, we examined the descriptives of the last sexual encounter for the whole sample as shown in Table 1. For the dependent variable, condom nonuse, 66.0% reported not using a condom during this last sexual encounter. In terms of timing, 92.9% reported that the sexual intercourse occurred in the last year and 7.1% reported that the sexual encounter did not occur in the last year. Next, differences between those who used a condom versus not at last sexual encounter were examined; results are shown in Table 1. Of note, substance use frequencies were reported rather than dependence diagnoses given the more descriptive nature of continuous data, but it is notable that the same findings (nonsignificant relationships) were obtained when dependence diagnoses were used instead. Those who did and did not use a condom did not differ significantly on the following environmental influences: history of childhood trauma, whether the partner injected drugs with a needle, was HIV positive, or was having intercourse with other people, or whether the participant was drunk or high during the last sexual intercourse. Those who did and did not use a condom did not differ significantly on the following psychopathology and affect variables: presence of a diagnosis of major depression, any anxiety disorder, BPD, or ASPD, negative emotionality, frequency of crack, heroin, alcohol, and marijuana use. Last, they also did not differ on HIV-related knowledge, or on the following self-regulatory skills/deficits: trait nonplanning impulsivity, delay discounting (all ps > .06).

Differences in Condom Use at Last Sexual Intercourse

Those who used a condom were significantly older than those who did not use a condom, t(154) = –1.99, p < .05. The CU versus nonuse groups significantly differed on the type of relationship with the partner, χ²(2, N = 156) = 11.03, p < .01. A larger percentage of those who did not use a condom described their partner as regular. Those who did not use a condom scored higher on risk-taking propensity as indexed by the BART, t(154) = –1.98, p < .05. And finally, those who did not use a condom reported more negative attitudes towards condoms then those who did use a condom, t(154) = –3.02, p < .01.

Finally, we conducted a multivariate logistic regression analysis to determine the independent contribution of each variable related to condom nonuse at a univariate level. Each independent variable found to have a significant univariate relationship to condom nonuse was included in a multivariate logistic regression analysis. Results are reported in Table 2. Overall, the model correctly classified 71.8% of the sample. Age did not remain significant (p > .05). Partner type, risk-taking propensity, and condom attitudes each were significant. Specifically, respondents were significantly less likely to report unprotected sex with casual partners, odds ratio (OR) = 0.25, 95% confidence interval (CI) [0.10, 0.60], p = .002, and commercial partners, OR = 0.26, 95% CI [0.09, 0.75], p = .01, compared to regular partners. For risk-taking propensity, a one unit increase in level of BART was associated with an approximately one and half greater likelihood of not having used a condom at last sexual intercourse, OR = 1.61, 95% CI [1.09, 2.34], p = .02. Finally, a one unit increase in negative condom attitudes was associated with an approximately one and three quarter greater likelihood of not having used a condom at last sexual intercourse, OR = 1.78, 95% CI [1.11, 2.85], p = .02.

Multivariate Logistic Regression Predicting Condom Nonuse at Last Sexual Intercourse

This study sought to examine variables associated with CU measured via event-level assessment among a sample of urban substance users. Utilizing the SAT as a guiding framework, a variety of SAT derived predictors were examined including: (1) environmental influences (i.e., childhood trauma and characteristics of last sexual intercourse), (2) forms of psychopathology and affect (diagnoses of depression, anxiety, BPD, and ASPD, substance use frequencies, and negative emotionality), (3) HIV-related attitudes and knowledge (condom attitudes and HIV knowledge), and (4) self-regulatory skills/deficits (trait nonplanning impulsivity, delay discounting, and risk-taking propensity). Findings from the logistic regression indicated that regular partner type, higher levels of risk taking propensity, and negative condom attitudes all contributed uniquely to an increased likelihood of condom nonuse.

These results represent one of the first attempts to apply the SAT to understanding the factors influencing RSB in an urban, largely minority, substance abusing sample utilizing event-level methodology. The selected independent variables represent a range of theoretically relevant factors related to RSB according to SAT, and the final model supports several key aspects of SAT theory, specifically how it posits the various dimensions contribute to health risk behavior and in this case RSB. Specifically, the significance of contextual, cognitive, and self-regulatory factors suggests the utility of SAT. Although the findings leave questions open for future exploration, the resulting model for RSB is conceptually meaningful, as it suggests that having sex with a regular partner, having negative attitudes towards condoms, and being prone to take risks combine to confer a relationship with engagement in RSB (i.e., condom nonuse). In the following sections we discuss the findings in the context of each SAT domain.

Within the domain of environmental influences, the finding that having sex with a regular partner was related to condom nonuse is consistent with prior work with nondrug using samples (e.g., Macaluso et al., 2000), but deserves additional attention in considering its meaning in the context of this study sample (urban substance users). In many samples, the choice to have intercourse without a condom when one's partner is regular may be thought of as a fairly low-risk behavior. However, condom nonuse even with a regular partner may pose elevated risk in this particular population given the prevalence of injection drug use, risky sexual behavior, and/or commercial sex in urban, substance using environments. Future studies would benefit from collecting more objective information on the regular partner, as this description may vary greatly among respondents, as well as determining the specific routines and situations in which sex with regular partners does or does not increase risk for condom nonuse. This finding supports the proposal by van Empelen and colleagues (2003) to develop programs that specifically target safe sexual behaviors of drug users in the steady relationship context.

Also from the domain of environmental influences, an additional notable finding was that being drunk and/or high during the last incident of sexual intercourse was not significantly related to CU. Although a common perception exists that alcohol and other substances negatively interfere with CU, a variety of studies have questioned this notion and posited that one's likelihood to use a condom may remain unchanged regardless of intoxication state (e.g., Leigh, Vanslyke, Hoppe, Rainey, Morrison, & Gillmore, 2008; Weinhardt & Carey, 2000). In a chronic substance using sample, disentangling the effect of substance use on CU is challenging, given that the pharmacological effects of different types of drugs have been shown to have different impacts on RSB (Leigh, Ames, & Stacy, 2008). In a recent study, only amphetamines (smoked or injected) were consistently related to decreased CU; alcohol use was not related to decreased CU, and cocaine and marijuana were not significantly related to CU in either direction (Leigh et al., 2008a). Thus, our finding that being drunk or high at the last sexual encounter was not related to CU may not be surprising when considered in the context of the existing literature, particularly given that the two most prevalent drug dependencies in our sample were cocaine (54.6%) and alcohol (34.3%). Another potential factor that may have limited a significant relationship between intoxication and CU is the chronic nature of the sample's substance use. For the individuals in our sample, intoxication or activities aimed at future substance use often are pervasive aspects of daily routines, and therefore isolating acute pharmacological effects on CU may be complicated.

Considering psychopathology and affect, no significant relationships with CU emerged, which is consistent with a meta-analysis reporting no significant correlation between depressive symptomatology, anxiety, anger, and sexual risk taking (Crepaz & Marks, 2001). However, the absence of findings also may have been overly influenced by the timing of assessment. That is, diagnoses certainly vary over time and may have in fact had been relevant at the time of the last sexual encounter. As with more global measures of RSB, this type of timing issue is a weakness of event-level measurement across a variety of variables. Nevertheless, it is likely that this limitation is most impactful for variables assessing psychopathology as compared to more stable variables such as attitudes and self-regulatory skills/deficits, suggesting the importance of future work considering other strategies to adequately assess the impact of these variables. It will be necessary to move beyond cross-sectional data and to examine within-person associations (Kalichman & Weinhardt, 2001). For example, using an online diary, Mustanski (2007) found that lower levels of positive affect are related to increased sexual risk behaviors. The diary method offers the advantage of a short time period between behavior and recording allowing for detailed information about the co-occurrence of events and moods that fits well with the internal and environmental aspects of SAT.

Within the domain of HIV-related attitudes and knowledge, condom attitudes predicted event-level RSB. This finding supports previous studies that have demonstrated the relevance of condom attitudes in predicting RSB (e.g., Robertson & Levin, 1999; Zamboni, Crawford, & Williams, 2000), and as such, intervention efforts that target condom attitudes as a means to increase safe sex practices (e.g., Jemmott, Jemmott, Braverman, & Fong, 2005; Wingood et al., 2006) may be useful to extend to urban substance using populations that face heightened HIV risk, with a particular focus on those with a propensity to take risks. The lack of significant relationship with HIV knowledge is consistent with prior research demonstrating that HIV/AIDS knowledge and perceived risk have little predictive value for safe sexual behavior (c.f. van Empelen et al., 2003). Findings for condom attitudes despite the absence of findings for HIV knowledge suggests that it is not necessarily what is known about the risks and consequences associated with HIV, but one's internal representation (attitudes in this case) resulting from knowledge and other external variables.

Finally, risk-taking propensity, as indexed by performance on the BART (a tendency that may limit self-regulatory resources) was related to condom nonuse during one's last sexual encounter. This is consistent with the Lejuez et al. (2004) study that found risk-taking propensity, as indexed by the BART, to be significant predictor of RSB, measured globally, above and beyond a host of variables. These findings serve as an extension of previously published reports and suggest the value of adding behavioral assessments such as the BART in understanding RSB, given that they are not compromised by limitations in individuals' ability to recognize and report on their own behavioral tendencies (Leigh & Stall, 1993; Tortu et al., 2000). Interpreting the factors underlying this relationship is difficult as the current study fails to rule out the possibility that both RSB as well as risk-taking propensity are the consequences of immersion in a substance using lifestyle as well as the direct consequence of neurotoxic illicit drugs, such as crack/cocaine (Booth, Kwiatkowski, & Chitwood, 2000; Hoffman et al., 2000; Ross et al., 2002). Further, this study fails to explain why trait nonplanning and delay discounting, both variables within the disinhibition umbrella, were not related to condom use as previous work among substance using samples has found impulsivity, measured with the Eysenck I7 Questionnaire (Eysenck, Pearson, Easting, & Allsopp, 1985), to be a statistically significant predictor of sexual risk after adjusting for the effects of demographic variables and substance use frequency (Hayaki et al., 2006). Future studies should be designed to systematically tease apart the potential contributions of dispositional factors from long-term negative effects of drug use in trying to better understand the origins of RSB. In addition, future work should also carefully consider the measurement of disinhibition as it is considered to be a multidimensional construct. It may be that certain aspects of disinhibition (e.g., risk-taking propensity) may be related to RSB whereas others are not. It is necessary to parse apart the differential findings with various measures of disinhibition, particularly the lack of consistency with self-report and behavioral tasks and the lack of significant correlation, as was found in this study, between measures that tap disinhibition.

The current study has a number of strengths, including the use of an at-risk, underserved sample of urban, primarily minority substance users, the application of the SAT framework, and the application of event-level measurement and behavioral assessment of risk-taking propensity to this population. However, several limitations should also be considered in interpreting the results. As noted above, there were multiple confounds associated with timing. It is unclear the extent to which measurement of study variables at our assessment accurately represent the manifestation of that variable at the most recent sexual experience, with only the contextual variables affecting condom use being variables in our battery directly tied to the most recent sexual encounter (i.e., to what extent are current condom attitudes at the assessment similar to those at the most recent sexual encounter; could not using a condom impact one's condom attitudes). Also presenting interpretative challenges, the current study did not assess the precise timing of the last sexual encounter (just whether the sexual encounter occurred in the previous year), thereby preventing statistical control for important factors (e.g., how long ago the last sexual encounter took place) to assess the temporal relationship among the predictors and our outcome measure.

Although the event-level measurement may help with overreporting and rounding and allows for an examination of situational variables, it is also important to acknowledge its limitations. There is the potential that the event reported may not be representative of the participant's overall behavior, and therefore cannot provide information on cumulative risk. The single event does not allow within person comparisons on behavior with regard to the environmental variables assessed, such as partner type or substance use, leaving open questions about whether the individual participant's behavior varies with regard to these factors. Further, the event-level report is still retrospective. Future work examining person-level predictors (e.g., risk taking propensity, condom attitudes) in addition to event characteristics (e.g., intoxication state, partner type) would benefit from repeated sampling with techniques as a daily diary to determine which variables are consistently associated with discrete behavior. Further, there is some evidence that recalls up to 90 days are reliable (e.g., Ajzen & Fishbein, 2004; Jaccard, McDonald, Wan, Dittus, & Quinlan, 2002); thus, in future work it may be best to limit event-level measurement to sexual encounters within the last 90 days or to conduct a prospective study that attempts to link current variables with sexual behavior.

Another limitation is the failure to assess sexual orientation and/or gender of sexual partner, a necessary target for future work. A final limitation is the size of the sample. Although sample size is acceptable for main findings, a number of interactions were examined (see footnote) that were not found to be significant. Although power was indeed limited for these analyses, it is notable that the odds ratios and confidence ratios indicate that these were very modest. Nevertheless, future studies would benefit form larger samples to further evaluate the utility of the SAT model, including examination of potential interaction effects.

The current data support a model framed in SAT in which having sex with a regular partner, having negative attitudes towards condoms, and being prone to take risks combine to confer a relationship with engagement in RSB (i.e., condom nonuse). Despite limitations and the questions raised by our findings, the current results add to the growing body of literature evaluating and predicting RSB within urban substance using populations and begins to lay the groundwork for future investigations that examine event-level RSB in the context of more elaborate environmental and self-regulatory predictors. By identifying personal and environmental self-regulatory resources that directly affect social interactions, problem solving, and self-efficacy, SAT applied to RSB has the potential to identity interventions that can be used to enhance health by altering self-goals, strategies, and environments. From a public health standpoint, identifying factors that influence HIV risk behaviors will continue to be critical, particularly as evidence grows suggesting that urban, minority substance users are at high risk for HIV infection.

¹ We also conducted several follow-up analyses. First, we examined if the relationship of condom use with risk taking propensity as well as condom attitudes varied by partner type. The interactions of partner type and risk taking propensity, OR = 1.00, 95% CI [0.97, 1.04], p = .94, and partner type and condom attitudes, OR = 1.36, 95% CI [0.71, 2.60], p = .36 were not significant. Next, although there was not a significant association between being drunk or high and condom use, an interaction between partner type and being drink or high was examined based on prior work demonstrating a significant interaction. The interaction of being drunk or high with partner type was not significant, OR = 2.05, 95% CI [0.66, 6.33], p = .21. Finally we examined gender interactions with the variables significantly related to condom use based on the univariate findings as shown in Table 1. There was not a significant interaction of gender with partner type, OR = 1.05, 95% CI [0.28, 3.93], p = .94, gender with condom attitudes, OR = 1.08, 95% CI [0.42, 2.75], p = .87, or gender with risk taking propensity, OR = 1.02, 95% CI [0.97, 1.07], p = .43.

Adimora, A. A., & Schoenbach, V. J. (2005). Social context, sexual networks, and racial disparities in rates of sexually transmitted infections. The Journal of Infectious Diseases, 191, S115–S122.

Ajzen, I., & Fishbein, M. (2004). Questions raised by a reasoned action approach: Comment on Ogden (2003). Health Psychology, 23, 431–434.

American Psychiatric Association. (1994). Diagnostic and statistical manual of mental disorders (4th ed.). Washington, DC: Author.

Arriola, K. R., Louden, T., Doldren, M. A., & Fortenberry, R. M. (2005). A meta-analysis of the relationship of child sexual abuse to HIV risk behavior among women. Child Abuse & Neglect, 29, 725–746.

Avants, S. K., Marcotte, D., Arnold, R., & Margolin, A. (2003). Spiritual beliefs, world assumptions, and HIV risk behavior among heroin and cocaine users. Psychology of Addictive Behaviors, 17, 159–162.

Bagozzi, R. P. (1992). The self-regulation of attitudes, intentions and behaviour. Social Psychology Quarterly, 55, 178–204.

Bandura, A. (1986). Social foundations of thought and action: A social cognitive. Englewood Cliffs, NJ: Prentice Hall.

Bandura, A. (1994). Social cognitive theory and exercise of control of HIV infection. In J.Peterson & R.DiClemente (Eds.), Preventing AIDS: Theory and practice of behavioral interventions (pp. 25–59). New York, NY: Plenum.

Bernstein, D. P., Fink, L., Handelsman, L., Foote, J., Lovejoy, M., Wenzel, K., . . .Ruggiero, J. (1994). Initial reliability and validity of a new retrospective measure of child abuse and neglect. The American Journal of Psychiatry, 151, 1132–1136.

Bernstein, D. P., Stein, J. A., & Newcomb, M. D. (2003). Development and validation of a brief screening version of the Childhood Trauma Questionnaire. Child Abuse & Neglect, 27, 169–190.

Booth, R. E., Kwiatkowski, C. F., & Chitwood, D. D. (2000). Sex related HIV risk behaviors: Differential risks among injection drug users, crack smokers, and injection drug users who smoke crack. Drug and Alcohol Dependence, 58, 219–26.

Brown, J. L., & Vanable, P. A. (2007). Alcohol use, partner type, and risky sexual behavior among college students: Findings from an event-level study. Addictive Behaviors, 32, 2940–2952.

Centers for Disease Control and Prevention. (2008). HIV prevalence estimates—United States. Morbidity and Mortality Weekly Report, 57, 1073–1076.

Chawarski, M. C., Pakes, J., & Schottenfeld, R. (1998). Assessment of HIV risk. Journal of Addictive Diseases, 17, 49–59.

Crepaz, N., & Marks, G. (2001). Are negative affective states associated with HIV sexual risk behaviors? A meta-analytic review. Health Psychology, 20, 291–299.

Ensminger, M. E., Anthony, J. C., & McCord, J. (1997). The inner city and drug use: Initial findings from an epidemiological study. Drug and Alcohol Dependence, 48, 175–184.

Ewart, C. K. (1991). Social action theory for a public health psychology. The American Psychologist, 46, 931–946.

Ewart, C. K. (2009). Changing our unhealthy ways: Emerging perspectives from Social Action Theory. In R.DiClemente, R.Crosby, & M.Kegler (Eds.), Emerging theories in health promotion practice and research (2nd ed.) (pp. 359–392). New York, NY: Jossey-Bass.

Eysenck, S. B. G., Pearson, P. R., Easting, G., & Allsopp, J. F. (1985). Age norms for impulsiveness, venturesomeness and empathy in adults. Personality and Individual Differences, 6, 613–619.

Fink, L. A., Bernstein, D., Handelsman, L., Foote, J., & Lovejoy, M. (1995). Initial reliability and validity of the Childhood Trauma Interview: A new multidimensional measure of childhood interpersonal trauma. The American Journal of Psychiatry, 152, 1329–1355.

First, M. B., Spitzer, R. L., Gibbon, M., & Williams, J. B. (2001). Structured Clinical Interview for DSM–IV–TR Axis I disorders, research version, non-patient edition (SCID–I/NP). New York, NY: Biometrics Research, New York State Psychiatric Institute.

Fisher, J. D., Fisher, W. A., Williams, S. S., & Malloy, T. E. (1994). Empirical tests of an information-motivation-behavioral skills model of AIDS-preventative behavior with gay men and heterosexual university students. Health Psychology, 13, 238–350.

Gillmore, M. R., Morrison, D. M., Leigh, B. C., Hoppe, M. J., Gaylord, J., & Rainey, D. T. (2002). Does “high = high risk”? An event-based analysis of the relationship between substance use and unprotected anal sex among gay and bisexual men. AIDS and Behavior, 6, 361–370.

Gollwitzer, P. M. (1990). Action phases and mind-sets. In. E. T.Higgins & R. M.Sorrentino (Eds.), Handbook of motivation and cognition: Foundations of social behavior (pp. 53–92). New York, NY: Guildford Press.

Gore-Felton, C., Rotheram-Borus, M. J., Weinhardt, L. S., Kelly, J. A., Lightfoot, M., Kirshenbaum, S. B., . . .NIMH Healthy Living Project Team. (2005). The Health Living Project: An individually tailored, multidimensional intervention for HIV-infected persons. AIDS Education and Prevention, 17, 21–39.

Hall, H. I., Song, R., Rhodes, P., Prejean, J., An, Q., Lee, L. M., . . .Janssen, R. S. (2008). Estimation of HIV incidence in the United States. Journal of the American Medical Association, 300, 520–529.

Harrison, J. D., Young, J. M., Butow, P., Salkeld, G., & Soloman, M. J. (2005). Is it worth the risk? A systematic review of instruments that measure risk propensity for use in the health setting. Social Science & Medicine, 60, 1385–1396.

Hayaki, J., Anderson, B., & Stein, M. (2006). Sexual risk behaviors among substance users: Relationship to impulsivity. Psychology of Addictive Behaviors, 20, 328–332.

Hoffman, J. A., Klein, H., Eber, M., & Crosby, H. (2000). Frequency and intensity of crack use as predictors of women's involvement in HIV-related sexual risk behaviors. Drug and Alcohol Dependence, 58, 227–236.

Ibañez, G. E., Van Oss Marin, B., Villarreal, C., & Gomez, C. A. (2005). Condom use at last sex among unmarried Latino men: An event-level analysis. AIDS and Behavior, 9, 433–441.

Jaccard, J., McDonald, R., Wan, C. K., Dittus, P. J., & Quinlan, S. (2002). The accuracy of self-reports of condom use and sexual behavior. Journal of Applied Social Psychology, 32, 1863–1905.

Jemmott, J. B., Jemmott, L. S., Braverman, P. K., & Fong, G. T. (2005). HIV/STD risk reduction interventions for African American and Latino adolescent girls at an adolescent medicine clinic: A randomized controlled trial. Archives of Pediatrics & Adolescent Medicine, 159, 440–449.

Johnson, M. O., Carrico, A. W., Chesney, M. A., & Morin. S. F. (2008). Internalized heterosexism among HIV-positive gay-identified men: Implications for HIV prevention and care. Journal of Consulting and Clinical Psychology, 76, 829–39.

Johnson, S. D., Cunningham-Williams, R. M., & Cottler, L. B. (2003). A tripartite of HIV-risk for African American women: The intersection of drug use, violence, and depression. Drug and Alcohol Dependence, 70, 169–175.

Kalichman, S. C., Rompa, D. L., & Webster, A. J. (2002). HIV transmission risk behaviours among HIV-positive persons in serodiscordant relationships. International Journal of STD & AIDS, 13, 677–682.

Kalichman, S. C., & Weinhardt, L. S. (2001). Negative affect and sexual risk behavior: Editorial comment. Health Psychology, 4, 300–301.

Kiene, S. M., Barta, W. D., Tennen, H., & Armeli, S. (2009). Alcohol, helping young adults have sex with casual partners: Findings from a daily diary study of alcohol use and sexual behavior. Journal of Adolescent Health, 44, 73–80.

Kirby, K. N., Petry, N. M., & Bickel, W. K. (1999). Heroin addicts have higher discount rates for delayed rewards than non-drug-using controls. Journal of Experimental Psychology: General, 128, 78–87.

LaBrie, J., Earleywine, M., Schiffman, J., Pederson, E., & Marriot, C. (2005). Effects of alcohol, expectancies, and partner type on condom use in college males: Event-level analyses. Journal of Sex Research, 42, 259–266.

Lehrer, J. A., Shrier, L. A., Gortmaker, S., & Buka, S. (2006). Depressive symptoms as a longitudinal predictor of sexual risk behaviors among US middle and high school students. Pediatrics, 118, 189–200.

Leigh, B., & Stall, R. (1993). Substance use and risky sexual behavior for exposure to HIV. Issues in methodology interpretation, and prevention. The American Psychologist, 48, 1035–1045.

Leigh, B. C., Ames, S. L., & Stacy, A. W. (2008a). Alcohol, drugs, and condom use among drug offenders: An event-based analysis. Drug and Alcohol Dependence, 93, 38–42.

Leigh, B. C., Vanslyke, J. G., Hoppe, M. J., Rainey, D. T., Morrison, D. M., & Gillmore, M. R. (2008b). Drinking and condom use: Results from an event-based daily diary. AIDS and Behavior, 12, 104–112.

Lejuez, C. W., Read, J. P., Kahler, C. W., Richards, J. B., Ramsey, S. E., Stuart, G. L., . . .Brown, R. A. (2002). Evaluation of a behavioral measure of risk-taking: The balloon analogue risk task (BART). Journal of Experimental Psychology: Applied, 8, 75–84.

Lejuez, C. W., Simmons, B. L., Aklin, W. M., Daughters, S. B., & Dvir, S. (2004). Risk-taking propensity and risky sexual behavior of individuals in residential substance use treatment. Addictive Behaviors, 29, 1643–1647.

Leonard, L., & Ross, M. W. (1997). The last sexual encounter: The contextualization of sexual risk behaviour. International Journal of STD & AIDS, 8, 643–645.

Lightfoot, M., Rotheram-Borus, M. J., Milburn, N. G., & Swendeman, D. (2005). Prevention for HIV-seropositive persons: Successive approximation toward a new identity. Behavior Modification, 28, 227–255.

Macaluso, M., Demand, M. J., Artz, L. M., & Hook, E. W. (2000). Partner type and condom use. AIDS, 14, 537–546.

Mazur, J. E. (1987). An adjusting procedure for studying delayed reinforcement. In M. L.Commons, J. E.Mazur, J. A.Nevin, H.Rachlin (Eds.), Quantitative analyses of behavior: Vol. 5. The effect of delay and of intervening events on reinforcement value (pp. 55–73). Hillsdale, NJ: Erlbaum.

McMahon, R. C., Malow, R. M., Devieux, J., Rosenberg, R., & Jennings, T. (2008). HIV risk and history of STDs in MCMI–III psychopathology subgroups of comorbid substance abusers. The American Journal of Drug and Alcohol Abuse, 34, 329–337.

Mellins, C. A., Dolezal, C., Brackis-Cott, E., Nicholson, O., & Meyer-Bahlburg, H. F. L. (2007). Predicting the onset of sexual and drug risk behaviors in HIV-negative youths with HIV-positive mothers: The role of contexual, self-regulation, and social-interaction factors. Journal of Youth and Adolescence, 36, 265–278.

Miller, M., & Neaigus, A. (2002). An economy of risk: Resource acquisition strategies of inner city women who use drugs. The International Journal on Drug Policy, 15, 409–418.

Mustanski, B. (2007). The influence of state and trait affect on HIV risk behaviors: A daily diary study of MSM. Health Psychology, 26, 618–626.

Palfai, T. P. (2006). Automatic processes in the self-regulation of addictive behaviors. In R. W.Wiers & A. W.Stacey (Eds.), Handbook of implicit cognition and addiction (pp. 411–424). Thousand Oaks, CA: Sage.

Patrick, C. J., Curtin, J. J., & Tellegen, A. (2002). Development and validation of a brief form of the Multidimensional Personality Questionnaire. Psychological Assessment, 14, 150–163.

Patton, J. H., Stanford, M. S., & Barratt, E. S. (1995). Factor structure of the Barratt Impulsiveness Scale. Journal of Clinical Psychology, 51, 768–774.

Rhodes, T., Singer, M., Bourgois, P., Friedman, S. R., & Strathdee, S. A. (2005). The social structural production of HIV risk among injecting drug users. Social Science & Medicine, 61, 1026–1044.

Robertson, A., & Levin, M. L. (1999). AIDS knowledge, condom attitudes, and risk-taking sexual behavior of substance-abusing juvenile offenders on probation or parole. AIDS Education and Prevention, 11, 450–461.

Ross, M. W., Hwang, L. Y., Zack, C., Bull, L., & Williams, M. L. (2002). Sexual risk behaviours and STIs in drug abuse treatment populations whose drug of choice is crack cocaine. International Journal of STD & AIDS, 13, 769–774.

Sanchez, J., Comerford, M., Chitwood, D. D., Fernandez, M. I., & McCoy, C. B. (2002). High risk sexual behaviours among heroin sniffers who have no history of injection drug use: Implications for HIV risk reduction. AIDS Care, 14, 391–398.

Saunders, J. B., Aasland, O. G., Babor, T. F., de la Fuente, J. R., & Grant, M. (1993). Development of the Alcohol Use Disorders Identification Test (AUDIT): WHO collaborative project on early detection of persons with harmful alcohol consumption–II. Addiction, 88, 791–804.

Scheidt, D. M., & Windle, M. (1996). Individual and situational markers of condom use and sex with nonprimary partners among alcoholic inpatients: Findings from the ATRISK study. Health Psychology, 15, 185–192.

Schroder, K. E., Carey, M. P., & Vanable, P. A. (2003). Methodological challenges in research on sexual risk behavior: II. Accuracy of self-reports. Annals of Behavioral Medicine, 26, 104–123.

Schwarzer, R. (1992). Self-efficacy in the adoption and maintenance of health behaviors: Theoretical approaches and a new model. In R.Schwarzer (Ed.), Self-efficacy: Thought control of action (pp. 217–243). Washington, DC: Hemisphere.

Semple, S. J., Grant, I., & Patterson, T. L. (2005). Negative self-perceptions and sexual risk behavior among heterosexual methamphetamine users. Substance Use & Misuse, 40, 1797–810.

Semple, S. J., Patterson, T. L., & Grant, I. (2000). HIV-positive gay and bisexual men: Predictors of unsafe sex. AIDS Care, 15, 3–15.

Senn, T. E., Carey, M. P., Vanable, P. A., Coury-Doniger, P., & Urban, M. A. (2006). Childhood sexual abuse and sexual risk behavior among men and women attending a sexually transmitted disease clinic. Journal of Consulting and Clinical Psychology, 74, 720–731.

Somlai, A. M., Kelly, J. A., Heckman, T. G., Hackl, K., Runge, L., & Wright, C. (2000). Life optimism, substance use, and AIDS-specific attitudes associated with HIV risk behavior among disadvantaged inner city women. Journal of Women's Health, 9, 1101–1111.

Somlai, A. M., Kelly, J. A., McAuliffe, T. L., Ksobiech, K., & Hackl, K. L. (2003). Predictors of HIV sexual risk behaviors in a community sample of injection drug-using men and women. AIDS and Behavior, 7, 383–393.

Tellegen, A. (1982). Brief manual for the Multidimensional Personality Questionnaire. Unpublished manuscript, University of Minnesota, Minneapolis.

Temple, M. T., & Leigh, B. C. (1992). Alcohol consumption and unsafe sexual behavior in discrete events. Journal of Sex Research, 29, 207–219.

Tortu, S., McMahon, J., Hamid, R., & Neaigus, A. (2000). Drug-using women's sexual risk: An event analysis. AIDS and Behavior, 4, 329–340.

van Empelen, P., Kok, G., van Kesteren, Ni. M. C., van den Borne, B., Bos, A. E. R., & Schaalma, H. P. (2003). Effective methods to change sex-risk among drug users: A review of psychosocial interventions. Social Science & Medicine, 57, 1593–1608.

Watkins, K. E., Metzger, D., Woody, G., & McLellan, A. T. (1993). Determinants of condom use among intravenous drug users. AIDS, 7, 719–723.

Weinhardt, L. S., & Carey, M. P. (2000). Does alcohol lead to sexual risk behavior? Findings from event-level research. Annual Review of Sex Research, 11, 125–157.

Wingood, G. M., DiClemente, R. J., Harrington, K. F., Lang, D. L., Davies, S. L., Hook, E. W., . . .Hardin, J. W. (2006). Efficacy of an HIV prevention program among female adolescents experiencing gender-based violence. American Journal of Public Health, 96, 1085–1090.

Zamboni, B. D., Crawford, I., & Williams, P. G. (2000). Examining communication and assertiveness as predictors of condom use: Implications for HIV prevention. AIDS Education and Prevention, 12, 492–504.

Submitted: April 14, 2009 Revised: October 26, 2009 Accepted: December 11, 2009

Assessing Motivational Interviewing Integrity for Group Interventions With Adolescents

Acknowledgement: We thank the Council on Alcoholism and Drug Abuse for their support of this project. We would also like to thank Cally Sprague and Susana Lopez for their help on the study. The current study was funded by a grant from the National Institute of Drug Abuse (R01DA019938) to Elizabeth D'Amico.

The group format is commonly used to treat adolescents who use alcohol and other drugs (AOD; Kaminer, 2005; Vaughn & Howard, 2004), even though little is known about the distinguishing factors of effective and ineffective group interventions (Engle, Macgowan, Wagner, & Amrhein, 2010). Recent work with at-risk adolescents indicates that motivational interviewing (MI) interventions (Miller & Rollnick, 2002; Rollnick, Miller, & Butler, 2008) can be successful, as MI offers a collaborative, nonjudgmental, and nonconfrontational approach (Naar-King & Suarez, 2010). Although studies of youth receiving individual MI have shown effectiveness (Baer, Garrett, Beadnell, Wells, & Peterson, 2007; D'Amico, Miles, Stern, & Meredith, 2008; Monti et al., 2007; Spirito et al., 2004), there are only four published group MI studies (Bailey, Baker, Webster, & Lewin, 2004; D'Amico, Osilla, & Hunter, 2010; Engle et al., 2010; Schmiege, Broaddus, Levin, & Bryan, 2009). Furthermore, MI treatment integrity has been mainly conducted with individual interventions (Baer et al., 2004; Martino, Ball, Nich, Frankforter, & Carroll, 2008; Moyers, Martin, Manuel, Hendrickson, & Miller, 2005; Vader, Walters, Prabhu, Houck, & Field, 2010). Only one study, to date, has evaluated MI integrity with adolescent group sessions (Engle et al., 2010) using measures of MI competence and adherence, such as the Motivational Interviewing Treatment Integrity scale (MITI; Moyers, Martin, Manuel, Miller, & Ernst, 2010).

Treatment integrity is important for ensuring that a therapy is delivered as intended by the treatment developers (Beidas & Kendall, 2010; Godley, Garner, Smith, Meyers, & Godley, 2011; Rakovshik & McManus, 2010). The MITI scale (Moyers et al., 2010) is a widely used measure of MI competence and adherence. Engle et al. (2010) used the MITI scale to examine the group process and found that group facilitator empathy was associated with more positive commitment language (e.g., “I'm quitting for the summer”), which was then associated with reduced AOD use (Engle et al., 2010). Although this study was pioneering, additional studies are needed, as only 19 sessions were coded, only one of the MITI global ratings (empathy) was assessed, and there was no comparison group. Thus, to move the field forward, more detailed treatment-integrity analyses (i.e., more group sessions and a comparison group) are needed so that we can assess how MI groups differ from other adolescent groups.

Finally, in order for treatment integrity to be successfully measured, alternative measurements to intraclass correlation (ICC) are needed to assess interrater agreement (e.g., Shrout & Fleiss, 1979). Calculating the ICC is not always feasible, as there may be different sets of raters for each session or there may be missing data (e.g., one session might be coded by all raters but another session may not). Thus, other approaches that compensate for these limitations must be utilized in order for studies to quantify interrater agreement.

The current study addresses these gaps by using the global and behavioral counts on the MITI to (a) assess agreement between raters, (b) calculate interrater agreement using two alternative methods to the ICC, and (c) evaluate the use of MI between a group led by facilitators trained and supervised in MI compared with a group with a self-help and Alcoholics Anonymous focus. This is an important first step in understanding whether the MITI is feasible for use in group settings and for determining MI integrity in this setting. We hypothesized that coders would reliably assess MI behaviors and that trained and supervised MI facilitators would deliver more MI-consistent behavior than the facilitator in the usual care (non-MI) group sessions.

Study Overview

This study was conducted as part of a randomized clinical trial (D'Amico, Hunter, et al., 2010). Procedures were approved by the research institution's internal review board. We collaborated with the Council on Alcoholism and Drug Abuse, a nonprofit organization in Santa Barbara, California, that operates a Teen Court (TC) for first-time offending youth. Based on earlier assessments, adolescents who do not need intensive treatment can participate in TC in lieu of formal juvenile justice processing. As part of TC, youth with a first time AOD offense are required to attend six AOD awareness groups, along with other sanctions (e.g., community service). Adolescents who successfully complete their sentence have the offense expunged from their record.

The AOD awareness groups occur weekly. Entrance to the groups is based on rolling admission, as each session can stand alone without a teen having to complete a previous session—that is, attending Session 2 does not require information from Session 1. Teens enter the sessions based on their time of referral to the program. Teens were randomized to either a usual care (UC) control condition or the experimental MI group intervention called Free Talk (FT). Teens were automatically assigned to attend the UC groups if they or their parent refused participation. Refusals (10%) were mostly due to lacking time or transportation to complete a baseline survey before their first group session. There were no demographic or offense differences between refusers and participants. See Table 1 for FT participant characteristics.

Characteristics of the Adolescents in the Free Talk and Usual Care Groups (N = 102)

The Current Study

This study focuses on the facilitator's behavior and whether this behavior can be reliably coded during adolescent group sessions. We provide detailed information on monitoring of a group MI intervention with at-risk adolescents (n = 102) across a large number of group sessions (n = 140) and compare these MI adolescent groups with usual care groups using the MITI.

Intervention Condition: Free Talk Groups

Free Talk was developed as part of a Stage I study (Rounsaville, Carroll, & Onken, 2001) where each group session was iteratively tested to determine feasibility and acceptability of intervention content (D'Amico, Osilla, et al., 2010). The facilitator manual for FT included a protocol for each session and utilized a MI approach. At the beginning of each session, the facilitator discussed the group guidelines (e.g., confidentiality, respect for others in the group). These were provided in a MI-consistent way (e.g., ensuring members were respectful of one another). FT is a manualized intervention that provides some pyschoeducation and focuses on encouraging change talk (see D'Amico, Osilla, et al., 2010, for content). For example, throughout all sessions, the focus is on providing reflections, asking open -ended questions to increase collaboration, affirming and supporting the adolescents to increase support and autonomy, and increasing change talk in the group by being aware of the DARN-C (Desire, Ability, Reason, Need, and Commitment; Amrhein, 2009). We also used tools to promote behavioral change, such as the decisional-balance activity and willingness and confidence rulers (Ingersoll, Wagner, & Gharib, 2006; Miller & Rollnick, 2002). Sessions lasted 55 min and mean group size was 4.5 adolescents (SD = 1.98).

Free Talk Training and Integrity Monitoring

The FT sessions were led by one of three facilitators (all female and White) who were psychology doctoral students with prior at-risk teen work experience. They received 40 hours of MI and FT training delivered by E.D. and K.O., clinical psychologists affiliated with the Motivational Interviewing Network of Trainers (MINT).

Facilitators were instructed to use MI to best fit a group format while still attending to individuals (D'Amico, Feldstein Ewing, et al., 2010; Velasquez, Stephens, & Ingersoll, 2006). For example, reflections were often stated to address the group process (e.g., “Many of you have been making positive changes in your lives”), and yet facilitators also responded to the varying individual experiences and needs (e.g., rolling with the resistance of one youth while trying to actively maintain the commitment language of another). In addition, facilitators monitored participant feedback and redirected negative or unhelpful comments to create a safe place for participation and mutual respect.

All FT groups were digitally audio recorded. MINT trainers reviewed all recordings and provided 1-hr weekly supervision to facilitators. The MITI was used to monitor performance (i.e., provide feedback during supervision).

Control Condition: Usual Care Groups

The UC condition was led by one facilitator who was male and Hispanic. The curriculum followed an abstinence-based Alcoholics Anonymous approach. Topics included group check-in, discussion of personal triggers, consequences of AOD use, educational videos, discussion of personal experiences with AOD use, and myths about AOD use. Like the FT groups, each session lasted about 55 min.

Coding

Overview

One hundred and 40 sessions (70 FT and 70 UC sessions) were coded by four psychology doctoral students. Per the MITI protocol, a randomly selected 20-minute segment was coded for both FT and UC sessions (Moyers et al., 2010). FT sessions were coded using digital audio recordings and UC sessions were coded from live observation, as not all teens attending these groups were study participants, so it was not possible to record these sessions. Raters were trained to not disrupt the UC groups by arriving early, sitting off to the side, not interacting with teens, and being discreet with coding materials.

MITI

The MITI is a widely used instrument for coding competency and adherence to MI and has been used in numerous studies to assess MI integrity (Jensen et al., 2011; Moyers et al., 2005; Tollison et al., 2008; Turrisi et al., 2009). It was specifically designed for clinical trials and is a reliable and valid tool (Madson & Campbell, 2006; Moyers et al., 2005). Integrity is measured through the use of global scores (e.g., amount of collaboration) and behavioral counts (e.g., number of reflections; Moyers et al., 2010). The MITI is used to assess the therapist's behavior during a session and focuses solely on the therapist's behavior; thus, the MITI can easily be utilized to assess a group facilitator's behavior (e.g., Engle et al., 2010). MITI 3.1 has five global scales (collaboration, empathy, evocation, autonomy/support, and direction) that are scored on a scale from 1 (low) to 5 (high). MITI competency is defined as an average of 4 on the global ratings (Moyers et al., 2010). As noted in the MITI manual, collaboration occurs when there is little power differential, there is agreement on goals, and clients are encouraged to share the talking. Empathy occurs when the facilitator expresses client understanding and attempts to understand client point of view. Evocation occurs when the facilitator encourages clients to brainstorm reasons and ideas for how to change. Autonomy/support occurs when the facilitator emphasizes and supports client's personal choice. Direction occurs when the facilitator exerts influence on the session and generally does not miss opportunities to direct client toward the target behavior or referral question (Moyers et al., 2010). In the group setting, the facilitator responds to group members, thus, group members influence the facilitator's behavior. For example, if a facilitator is not collaborative and does not ask open-ended questions, then there will be less participation and less sharing of the talking. Facilitator responsiveness to group member behavior is captured by the MITI global scores.

The rater also counts specific behaviors that occur during each coded segment, including open-ended questions, closed-ended questions, MI-adherent (e.g., “If it's ok with you, I'd like to share some information with you”) and nonadherent statements (e.g., “You need to stop drinking”), and simple (e.g., “some of you are ready to make changes”) and complex reflections (e.g., “some of you are hoping that by making changes, things will improve in your lives”). Whereas global scores have a range limit (1 to 5), behavioral counts have no upper end on the scale; thus, these scores can vary by session to a greater degree.

Coding Training

Four raters received about 40 hours of training, which included a half-day MITI training and practice coding assignments (http://casaa.unm.edu/codinginst.html). Similar to other studies (Moyers et al., 2005; Tollison et al., 2008), raters met weekly to discuss discrepancies. Interrater agreement was stable over time. Raters coded both UC and FT group sessions.

Interrater Agreement for MITI

All 140 sessions were coded by at least one rater, with 48 (34%) sessions coded by two raters (19 UC and 29 FT) and 25 (18%) FT sessions coded by three raters. We did not have three raters code UC sessions, given that these groups were coded “live” and this would have been disruptive. Raters were not the same two people each time, as they were assigned to code sessions randomly. Thus, the ICC could not be calculated, as different raters coded different sessions and ICC is not defined in the presence of missing data. In addition, using the ICC in this case would not be an accurate comparison, as it is like reporting correlations among items rather than a coefficient alpha. Interpretation of these coefficients would also be difficult because many of them have small sample sizes and hence we do not have a sense of the sampling distribution of the statistic.

We therefore used two different methods to quantify agreement between raters. Our first method was distribution-free: We calculated the difference between each rater's rating of a session and the mean rating for that session. To see how close the majority of ratings were to the mean, we ordered these differences from the smallest to the largest. We then took the 95th percentile (e.g., if there were 100 differences, we took the 95th rating, referred to as D95 in Table 2). Thus, the D95 value indicates that 95% of differences between each rater and the mean rating are smaller than this value. The D95 is more appropriate for dichotomous variables, as it makes no distributional assumptions.

MITI Measures of Agreement Between Raters and Mean Scores by Rater Across Both the Free Talk and Usual Care Groups

In our second method, we used a normal distribution approach of calculating the within-session standard deviation (WSSD) to provide an estimate of the difference between raters (Altman & Bland, 1983; Bland & Altman, 1986, 2007). The WSSD is closely related to the average of the difference between raters, ∑(x−
 /N, where
  is the mean for the session and x represents the set of ratings. The WSSD is given by the formula for the standard deviation (∑(x−
 )²/(N−1))^0.5. Because the WSSD squares the differences between each rating and the mean, larger deviations between raters lead to proportionally larger estimates. If we assume differences are normally distributed, we expect 95% of ratings to lie within about two standard deviations of that mean. For interpretation purposes, we therefore report 2*WSSD, which allows us to compare the two methods.

These two methods make different assumptions about the distribution of ratings. The D95 method does not make distributional assumptions, but it does not use all of the information that is available regarding the values. In contrast, the WSSD method uses all the information and assumes that data are normally distributed and measured on a continuous scale. Using two methods allowed us to examine interrater agreement sensitivity and compare the results of the two methods.

These agreement measures have two advantages over the ICC. First, the ICC is a ratio of between-group variance to total variance. The ICC depends not only on the extent of the agreement between raters but also on the variance of the measures. Thus, the ICC could be lower or higher, depending on the level of variance, even if the rater agreement level was the same in both cases. Second, the ICC presents agreement on a scale from 0.0 (no agreement) to 1.0 (perfect agreement). Although the scale allows for a comparison across studies, the ICC has little interpretive value of how actual scores differ between raters, as the units of the ICC are not the same as the units on the scale. This variation is very important for scales like the MITI that have items ranging from 1 to 5 points because it provides data on how to recalibrate coding to ensure that they match.

Interrater Agreement

We calculated the limits of agreement between the raters for each of the MITI dimensions using the D95 and 2*WSSD methods. Table 2 shows ratings for both groups combined across all four raters. For MITI global ratings using the D95 method, raters were within 0.5 points of the mean for the session 95% of the time. Using the 2*WSSD method, the level of agreement was similar, with expected 95% limits ranging from 0.50 (autonomy/support and empathy) to 0.62 (evocation).

For the behavioral counts, there was no upper limit; thus, we expected to see larger differences. Both approaches yielded comparable limits of agreement. The D95 limits ranged from 3.56 (complex reflections) to 5.56 (MI-adherent and closed questions). The 2*WSSD limits ranged from 3.26 (complex reflections) to 5.54 (closed questions). As one would expect, MI-nonadherent counts had lower D95 (0.50) and 2*WSSD (0.36). There are no published MITI benchmarks for what consists of good agreement. For this study, we consider within 1 point on the MITI global scores and within 6 points on the behavioral counts as good levels of agreement.

MI Differences by Group

Table 3 shows differences between the means of all raters across groups. Overall, both global and behavioral count scores were higher for the FT group compared with the UC group, indicating greater MI integrity in the FT group sessions.

MITI Means and Standard Deviations Across Raters by Group

This study is one of the first to assess MI integrity using data from an adolescent group intervention. First, this study found that group MI can be efficiently and reliably assessed using alternatives to ICC. We used two innovative methods to address agreement and both found comparable results. These methods are more useful than the ICC because they can be calculated when using multiple coders and are better able to quantify the difference between multiple raters, providing us with a more direct interpretation of the level of agreement. For example, a D95 or 2*WSSD score of 2 tells us that 95% of raters are expected to be within 2 points of the mean of all raters. A clinical judgment must then be made to determine whether the raters are sufficiently close to agreement, which depends upon on the units of the original measure. A difference of 2 points on global ratings is very large compared with behavioral counts. Thus, our findings of less than .62 difference between a rater and the mean score on the global ratings and 5.56 on the behavioral counts suggest that the raters were in fairly close agreement. Using these indices of interrater agreement could be useful in clinical practice, as it bridges the gap between research and practice by allowing for real-world situations such as multiple raters or supervisors.

We found that it is feasible to train facilitators to conduct MI in an adolescent group setting and that MI can be measured and implemented with integrity. Overall, global scores for the FT group were in the competent range and behavioral counts for MI behaviors were high. We expected this, given the intensive training and supervision; MI fidelity might not be as high with less supervision. In addition, future research could measure integrity across more facilitators.

We found large differences between the FT and UC groups on collaboration, empathy, evocation, and autonomy/support, suggesting that FT groups were more likely to encourage power sharing during the session, had repeated efforts to gain understanding of teens' viewpoints, had more acceptance of teens' reasons for change, and were more likely to emphasize support of client autonomy. The behavioral count data showed a similar pattern with more simple and complex reflections and open-ended questions in FT groups. More MI-nonadherent behavior was observed in the UC groups, such as confronting and advising without permission compared with the FT groups. These findings are not surprising, given that the UC group facilitator did not receive MI training or supervision but rather followed an Alcoholics Anonymous group treatment approach. Our study findings suggest that, similar to MI research with individuals (Moyers et al., 2005), group MI could make the overall group process more collaborative and therefore more effective, and this may reduce the likelihood of iatrogenic effects that are sometimes seen in groups of at-risk youth (e.g., Dishion, McCord, & Poulin, 1999; Dodge, Dishion, & Lansford, 2006). For example, as shown with individual work (Magill, Apodaca, Barnett, & Monti, 2010; Moyers, Martin, Houck, Christopher, & Tonigan, 2009), the increased evocation and autonomy/support may provide adolescents with a safe place to explore reasons for change without fear of being forced to change; the guiding and empathic style of MI may then help move participants toward positive behavior change.

This study had limitations. First, raters were not blind to study condition. Similar to other MI brief interventions (e.g., Hettema, Steele, & Miller, 2005), FT sessions followed a protocol and could be easily distinguished from other types of sessions. Second, our UC control condition had to be coded by live observation; thus, it is possible that some information was missed. However, raters coded both UC and FT with one pass, so we feel confident that raters captured most of the behaviors during the UC sessions. Moreover, our results indicated that raters were consistent in their ratings across both groups and had high agreement, suggesting that they were not biased. Third, no prior research has employed these statistical methods to address interrater agreement for the MITI; thus, no consensus exists regarding how much difference among raters might be “acceptable.” We do not consider this a problem, as cutoff values may vary across studies and may not be quantifiable, and the use of such values could limit researchers from interpreting results objectively. Thus, it is important to use clinical judgment when evaluating the level of agreement with these methods. Finally, we did not assess group cohesion or engagement; more work is needed in this area.

In sum, this study takes an important first step in documenting group MI with at-risk youth. This study is among the first to provide within-session standard deviations for all global and behavioral counts that comprise the MITI scoring system. Thus, this work can provide a potential resource to practitioners and researchers who do group work and are interested in using MI and measuring MI integrity in this setting. The next step in this work is to examine intervention efficacy; we are currently conducting a randomized trial to assess the short-term effects of this intervention on AOD use among at-risk teens.

Altman, D. G., & Bland, J. M. (1983). Measurement in medicine: The analysis of method comparison studies. The Statistician, 32, 307–317. doi:10.2307/2987937

Amrhein, P. C. (2009). A training manual for coding client commitment language. New York, NY: Columbia University.

Baer, J. S., Garrett, S. B., Beadnell, B., Wells, E. A., & Peterson, P. L. (2007). Brief motivational intervention with homeless adolescents: Evaluating effects on substance use and service utilization. Psychology of Addictive Behaviors, 21, 582–586. doi:10.1037/0893-164X.21.4.582

Baer, J. S., Roesengren, D. B., Dunn, C. W., Wells, E. A., Ogle, R. L., & Hartzler, B. (2004). An evaluation of workshop training in motivational interviewing for addiction and mental health clinicians. Drug and Alcohol Dependence, 73, 99–106. doi:10.1016/j.drugalcdep.2003.10.001

Bailey, K. A., Baker, A. L., Webster, R. A., & Lewin, T. J. (2004). Pilot randomized controlled trial of a brief alcohol intervention group for adolescents. Drug and Alcohol Review, 23, 157–166. doi:10.1080/09595230410001704136

Beidas, R. S., & Kendall, P. C. (2010). Training therapists in evidence-based practice: A critical review of studies from a systems-contextual perspective. Clinical Psychology: Science and Practice, 17, 1–30. doi:10.1111/j.1468-2850.2009.01187.x

Bland, J. M., & Altman, D. G. (1986). Statistical methods for assessing agreement between two methods of clinical measurement. Lancet, 1, 307–310. doi:10.1016/S0140-6736(86)90837-8

Bland, J. M., & Altman, D. G. (2007). Agreement between methods of measurement with multiple observations per individual. Journal of Biopharmaceutical Statistics, 17, 571–582. doi:10.1080/10543400701329422

D'Amico, E. J., Feldstein Ewing, S. W., Engle, B., Hunter, S. B., Osilla, K. C., & Bryan, A. (2010). Group alcohol and drug treatment. In S.Naar-King & M.Suarez (Eds.), Motivational interviewing with adolescents and young adults (pp. 151–157). New York, NY: Guilford Press.

D'Amico, E. J., Hunter, S. B., Osilla, K. C., Miles, J. N. V., Munjas, B., & Saunders, J. (2010). Group MI for first time alcohol and marijuana offending adolescents. Paper presented at the Research Society on Alcoholism, San Antonio, TX.

D'Amico, E. J., Miles, J. N. V., Stern, S. A., & Meredith, L. S. (2008). Brief motivational interviewing for teens at risk of substance use consequences: A randomized pilot study in a primary care clinic. Journal of Substance Abuse Treatment, 35, 53–61. doi:10.1016/j.jsat.2007.08.008

D'Amico, E. J., Osilla, K. C., & Hunter, S. B. (2010). Developing a group motivational interviewing intervention for adolescents at-risk for developing an alcohol or drug use disorder. Alcoholism Treatment Quarterly, 28, 417–436. doi:10.1080/07347324.2010.511076

Dishion, T. J., McCord, J., & Poulin, F. (1999). When interventions harm: Peer groups and problem behavior. American Psychologist, 54, 755–764. doi:10.1037/0003-066X.54.9.755

Dodge, K., Dishion, T. J., & Lansford, J. E. (2006). Deviant peer influences in programs for youth. New York, NY: Guilford Press.

Engle, B., Macgowan, M. J., Wagner, E. F., & Amrhein, P. (2010). Markers of marijuana use outcomes within adolescent substance abuse group treatment. Research on Social Work Practice, 20, 271–282. doi:10.1177/1049731509347855

Godley, S. H., Garner, B. R., Smith, J. E., Meyers, R. J., & Godley, M. D. (2011). A large-scale dissemination and implementation model for evidence-based treatment and continuing care. Clinical Psychology: Science and Practice, 18, 67–83. doi:10.1111/j.1468-2850.2011.01236.x

Hettema, J., Steele, J., & Miller, W. R. (2005). Motivational interviewing. Annual Review of Clinical Psychology, 1, 91–111. doi:10.1146/annurev.clinpsy.1.102803.143833

Ingersoll, K. S., Wagner, C. C., & Gharib, S. (2006). Motivational groups for community substance abuse programs (3rd ed.). Rockville, MD: Substance Abuse Mental Health Services Administration.

Jensen, C. D., Cushing, C. C., Aylward, B. S., Craig, J. T., Sorell, D. M., & Steele, R. G. (2011). Effectiveness of motivational interviewing interventions for adolescent substance use behavior change: A meta-analytic review. Journal of Consulting and Clinical Psychology, Online First Publication, 79, 433–440.

Kaminer, Y. (2005). Challenges and opportunities of group therapy for adolescent substance abuse: A critical review. Addictive Behaviors, 30, 1765–1774. doi:10.1016/j.addbeh.2005.07.002

Madson, M. B., & Campbell, T. C. (2006). Measures of fidelity in motivational enhancement: A systematic review. Journal of Substance Abuse Treatment, 31, 67–73. doi:10.1016/j.jsat.2006.03.010

Magill, M., Apodaca, T. R., Barnett, N. P., & Monti, P. M. (2010). The route to change: Within-session predictors of change plan completion in a motivational interview. Journal of Substance Abuse Treatment, 38, 299–305. doi:10.1016/j.jsat.2009.12.001

Martino, S., Ball, S. A., Nich, C., Frankforter, T. L., & Carroll, K. M. (2008). Community program therapist adherence and competence in motivational enhancement therapy. Drug and Alcohol Dependence, 96, 37–48. doi:10.1016/j.drugalcdep.2008.01.020

Miller, W. R., & Rollnick, S. (2002). Motivational interviewing: Preparing people for change (2nd ed.). New York, NY: Guilford Press.

Monti, P. M., Barnett, N. P., Colby, S. M., Gwaltney, C. J., Spirito, A., Rohsenow, D. J., …Woolard, R. (2007). Motivational interviewing versus feedback only in emergency care for young adult problem drinking. Addiction, 102, 1234–1243. doi:10.1111/j.1360-0443.2007.01878.x

Moyers, T. B., Martin, T., Houck, J. M., Christopher, P. J., & Tonigan, J. S. (2009). From in-session behaviors to drinking outcomes: A causal chain for motivational interviewing. Journal of Consulting and Clinical Psychology, 77, 1113–1124. doi:10.1037/a0017189

Moyers, T. B., Martin, T., Manuel, J. K., Hendrickson, S. M. L., & Miller, W. R. (2005). Assessing competence in the use of motivational interviewing. Journal of Substance Abuse Treatment, 28, 19–26. doi:10.1016/j.jsat.2004.11.001

Moyers, T. B., Martin, T., Manuel, J. K., Miller, W. R., & Ernst, D. (2010). Revised Global Scales: Motivational Interviewing Treatment Integrity 3.1.1 (MITI 3.1.1). Retrieved from http://casaa.unm.edu/mimanuals.html

Naar-King, S., & Suarez, M. (2010). Motivational interviewing with adolescents and young adults. New York, NY: Guilford Press.

Rakovshik, S. G., & McManus, F. (2010). Establishing evidence-based training in cognitive behavioral therapy: A review of current empirical findings and theoretical guidance. Clinical Psychology Review, 30, 496–516. doi:10.1016/j.cpr.2010.03.004

Rollnick, S., Miller, W. R., & Butler, C. C. (2008). Motivational interviewing in health care: Helping patients change behaviorNew York, NY: Guilford Press.

Rounsaville, B. J., Carroll, K. M., & Onken, L. S. (2001). A stage model of behavioral therapies research: Getting started and moving on from stage I. Clinical Psychology: Science and Practice, 8, 133–142. doi:10.1093/clipsy.8.2.133

Schmiege, S. J., Broaddus, M. R., Levin, M., & Bryan, A. D. (2009). Randomized trial of group interventions to reduce HIV/STD risk and change theoretical mediators among detained adolescents. Journal of Consulting and Clinical Psychology, 77, 38–50. doi:10.1037/a0014513

Shrout, P. E., & Fleiss, J. L. (1979). Intraclass correlations: Uses in assessing rater reliability. Psychological Bulletin, 86, 420–428. doi:10.1037/0033-2909.86.2.420

Spirito, A., Monti, P. M., Barnett, N. P., Colby, S. M., Sindelar, H., Rohsenow, D., …Myers, M. (2004). A randomized clinical trial of a brief motivational intervention for alcohol-positive adolescents treated in an emergency department. Journal of Pediatrics, 145, 396–402. doi:10.1016/j.jpeds.2004.04.057

Tollison, S. J., Lee, C., Neighbors, C., Neil, T., Olson, N., & Larimer, M. (2008). Questions and reflections: The use of motivational interviewing microskills in a peer-led brief alcohol intervention for college students. Behavior Therapy, 39, 183–194. doi:10.1016/j.beth.2007.07.001

Turrisi, R., Larimer, M., Mallett, K., Kilmer, J., Ray, A., Mastroleo, N., …Montoya, H. (2009). A randomized clinical trial evaluating a combined alcohol intervention for high-risk college students. Journal of Studies on Alcohol and Drugs, 70, 555–567.

Vader, A. M., Walters, S. T., Prabhu, G. C., Houck, J. M., & Field, C. A. (2010). The language of motivational interviewing and feedback: Counselor language, client language, and client drinking outcomes. Psychology of Addictive Behaviors, 24, 190–197. doi:10.1037/a0018749

Vaughn, M. G., & Howard, M. O. (2004). Adolescent substance abuse treatment: A synthesis of controlled evaluations. Research on Social Work Practice, 14, 325–335. doi:10.1177/1049731504265834

Velasquez, M. M., Stephens, N. S., & Ingersoll, K. S. (2006). Motivational interviewing in groups. Journal of Groups in Addiction and Recovery, 1, 27–50. doi:10.1300/J384v01n01_03

Submitted: August 1, 2011 Revised: November 8, 2011 Accepted: January 9, 2012

Association of Solitary Binge Drinking and Suicidal Behavior Among Emerging Adult College Students

Acknowledgement: See page 620 for a correction to this article.

This research was supported by funds provided through the University of Alaska Anchorage Chancellor's Fund and a National Institute on Alcohol Abuse and Alcoholism grant (R21AA018135) to Vivian M. Gonzalez.

College students have high rates of “binge” drinking, defined as four or more drinks for women or five or more drinks for men in one sitting or on one occasion (Wechsler, Davenport, Dowdall, & Moeykens, 1994), and alcohol problems (Knight et al., 2002; Slutske, 2005; Timberlake et al., 2007). American college students also engage in high rates of suicidal ideation and behavior (Brener, Hassan, & Barrios, 1999; Furr, Westefeld, McConnell, & Jenkins, 2001), with 1.7% reporting a suicide attempt and 1.4% reporting seriously having considered attempting suicide during the preceding year (Barrios, Everett, Simon, & Brener, 2000).

There is a well-established link between alcohol use and disorders and suicide attempts and completions (Borges, Walters, & Kessler, 2000; Cherpitel, Borges, & Wilcox, 2004; Hufford, 2001; Wilcox, Conner, & Caine, 2004). Among college students, individuals with suicidal ideation are more likely to binge drink, and alcohol problems in this population are associated with increased rates of suicidal ideation and attempts (Brener et al., 1999; Gonzalez, Bradizza, & Collins, 2009; Levy & Deykin, 1989; Stephenson, Pena-Shaff, & Quirk, 2006).

Research suggests that the drinking context plays an important role in the association between alcohol use and negative affect, including suicidality. In college students, drinking tends to occur for social reasons (Kuntsche, Knibbe, Gmel, & Engels, 2005; LaBrie, Hummer, & Pedersen, 2007; Stewart, Zeitlin, & Samoluk, 1996) and in social contexts (Christiansen, Vik, & Jarchow, 2002; Mohr et al., 2001). However, drinking in response to negative experiences and affect is associated with drinking in solitary, as opposed to social, contexts (Mohr et al., 2001). Students who engage in solitary binge drinking (i.e., while alone or when no one else is drinking) have more severe depression (Christiansen et al., 2002) and suicidal ideation (Gonzalez, Collins, & Bradizza, 2009) than students who only binge drink in social contexts.

Depression and suicidal ideation are associated with suicide and suicide attempts (Brown, Beck, Steer, & Grisham, 2000; Kuo, Gallo, & Tien, 2001; Kuo, Gallo, & Eaton, 2004). Given the association of solitary binge drinking with these variables, solitary binge drinkers may be at higher risk for suicidal behavior than are students who binge drink only in social contexts. However, no study to date has examined the association between solitary drinking and previous suicide attempts, one of the best predictors of risk for suicide and suicide attempts (Borges, Angst, Nock, Ruscio, & Kessler, 2008; Harris & Barraclough, 1997; Oquendo et al., 2004).

This study examined whether individuals with a suicide attempt history were more likely to be solitary as opposed to social binge drinkers. The associations of suicide attempt history and suicidal ideation with frequency and quantity of drinking in social and solitary contexts also were explored to further examine the association of drinking context and suicidality.

Participants

Participants for the current study were drawn from two studies of emerging adult college drinkers (Gonzalez & Skewes, 2011; Gonzalez, Reynolds, & Skewes, 2011). For both studies, eligibility criteria included (a) being a full- or part-time university student and (b) being between the ages of 18 and 25 years old. Each study had additional selection criteria. In Study 1, participants had to have at least one solitary or social binge drinking episode per month during a typical month in the past year. In Study 2, participants had to have consumed at least four standard drinks in the past month and report either current sadness or loss of pleasure. The sadness and loss of pleasure items were adapted from the Beck Depression Inventory-II (Beck, Steer, & Brown, 1996), with items needing to be endorsed as at least a 1 (i.e., not enjoying things as much or feeling sad much of the time) to meet the inclusionary criterion. Although Study 2 included the sadness or loss of pleasure criterion, combining these samples was advantageous for the current study, as depression has been found to be associated with the key variables of interest in this study: solitary binge drinking and suicidality. Data in these studies were collected from March 2009 to January 2010. For the current study, individuals were not included if they did not binge drink during a typical month in the past year (n = 31), gave out-of-range and/or highly inconsistent and illogical responses across alcohol use items (n = 25), or were missing drinking data (n = 2).

Participants were 182 binge drinking, emerging adult (18- to 25-year-old) female (72.5%, n = 132) and male (27.5%, n = 50) college students attending a large, open-enrollment university in the Northwest. The average age was 21.1 years old (SD = 1.9). The sample was 74.7% White, 10.4% Alaska Native or American Indian (including mixed heritage), 4.4% Latino, 4.4% multiethnic, 2.2% Asian American, 1.6% Pacific Islander, and 1.1% African American. The majority of participants were full-time students (88.5%) and single (95.6%). With regard to living arrangements, 23.8% lived in their parents' home, 53.5% lived off campus (not in a parent's home), and 22.7% lived on campus. The sample was 19.8% freshman, 24.2% sophomore, 24.2% junior, 23.6% senior, 6.0% graduate students, and 2.2% nondegree-seeking.

Procedures

In both Studies 1 and 2, participants were recruited via flyers posted on campus and e-mails directed at 18- to 25-year-old students via their student e-mail accounts. Flyer and e-mail solicitations directed potential participants to a webpage that described the respective study in general terms (e.g., “a study of college student lifestyle and mood”) and included a questionnaire with items that screened for study eligibility embedded among distractor items. Those who met eligibility criteria for each respective study were scheduled for an in-person data collection session where study materials were presented in random order using MediaLab version 2006 software (Jarvis, 2006) on laptop computers. Participants were compensated for their time with a gift card ($20 for Study 1 and $30 for Study 2, given the longer protocol) to a supermarket/gas station or coffee-shop chain. After data collection, all participants were individually debriefed and given referral information for counseling services, as well as suicide/crisis-hotline phone numbers. The study protocols were approved by the institutional review board of the university.

Measures

Alcohol consumption

Solitary and social alcohol consumption were measured using items modified from the National Institute on Alcohol Abuse and Alcoholism's (NIAAA) alcohol consumption question set (NIAAA, 2003). Three separate items for social (defined as “with other people who were drinking”) and solitary (defined as “while alone or no one else was drinking”) contexts asked the following for a typical month in the past year: drinking days per month, number of standard drinks consumed on a typical drinking day, and the number of days on which binge drinking (i.e., four or more for women, or five or more for men, standard drinks on one occasion or sitting) occurred. Participants were provided with a handout that defined a standard drink (e.g., 12 oz. of beer, 5 oz. of wine, 8 to 9 oz. of malt liquor, or 1.5 oz. of 80-proof liquor).

Suicide attempt history

The Suicidal Behaviors Questionnaire—Revised (SBQ-R; Osman et al., 2001) is a 4-item self-report measure of suicidal behavior and ideation. The first item assesses lifetime suicidal ideation and behavior (“Have you ever thought about or attempted to kill yourself?”), with mutually exclusive response options ranging from no history of suicidal ideation or behavior (never) to history of a suicide attempt (I have attempted to kill myself). This item was used in the current study to categorize participants as having or not having a suicide attempt history. In young adult samples, the SBQ-R demonstrates high 2-week test–retest reliability (r = .95) and good convergent validity (Cotton, Peters, & Range, 1995).

Suicidal ideation

The Adult Suicidal Ideation Questionnaire (ASIQ; Reynolds, 1991a) is a 25-item self-report measure of suicidal thoughts and behavior experienced during the past month. Items range from general wishes that one were dead to thoughts of planning a suicide attempt and are rated on a 7-point scale (0 = never had the thought to 6 = almost every day). The ASIQ demonstrates high 1-week test–retest reliability (r = .86) and good convergent validity (Gutierrez, Osman, Kopper, Barrios, & Bagge, 2000; Reynolds, 1991b). It also evidences predictive validity, with total score predicting suicide attempts over a 3-month follow-up period (Osman et al., 1999). In the current sample, coefficient alpha for the ASIQ was .97.

Analyses

A hierarchal logistic regression was used to examine the influence of a previous suicide attempt and severity of suicidal ideation on binge drinking group (social only = 0, solitary = 1). For these analyses, if a participant reported (a) no episodes of solitary binge drinking during a typical month in the past year, and (b) had at least one episode of social binge drinking, then they were classified as a social binge drinker (n = 129). If a participant reported a binge drinking episode while alone or when no one else was drinking at least once during a typical month in the past year, then they were classified as a solitary binge drinker (n = 53). Solitary binge drinkers also could have episodes of social binge drinking, and previous research has found that nearly all do engage in social binge drinking (Gonzalez, Collins et al., 2009).

Four separate hierarchical multiple regression analyses for frequency of drinking and drinks per drinking day in social and solitary contexts were conducted to examine the association of a previous suicide attempt and suicidal ideation with each drinking variable. The same independent variables were entered in the multiple regression analyses and logistic regression analysis described above. In the first step of the analyses, age, gender, and ethnicity were entered to control for their possible effects on drinking and suicidality variables. In the second step, suicide attempt history (no attempts = 0, previous attempt = 1) was entered. In the final step, severity of suicidal ideation was entered to examine whether suicidal ideation was significantly associated with the given drinking dependent variable after accounting for suicide attempt history.

Given that the data for the current research came from two studies, analyses were repeated, including data source (Study 1 vs. 2) in the models in Step 1 and mean-centered interaction terms of Data Source × Suicide Attempt History and Data Source × Suicidal Ideation included in a final step. The interaction terms allowed an examination of whether the degree of relationship between suicide attempt history and severity of suicidal ideation and the dependent variables differed between the samples in the two studies.

In order to improve normality and reduce the influence of outliers, suicidal ideation, frequency of social drinking, and drinks per social drinking day were square-root transformed. Frequency of solitary drinking and drinks per solitary drinking day, which were more substantially skewed, were log transformed.

In the study sample, 22.6% (n = 12) of solitary binge drinkers and 8.5% (n = 11) of social binge drinkers reported a previous suicide attempt. Means, standard deviations, and correlations among the study variables are presented in Table 1. The majority of solitary binge drinkers also engaged in social binge drinking at least one day per month (94.3%). An analysis of covariance, controlling for age, gender, and ethnicity revealed that solitary binge drinkers engage in social binge drinking more days per month (M = 5.03, SD = 3.98) than individuals who only binge drink socially (M = 3.38, SD = 2.81; F(1, 177) = 10.01, p = .002, η² = .05). Solitary binge drinkers reported a mean of 3.04 (SD = 3.97) solitary binge days per month. Among social binge drinkers, 50.4% engaged in solitary drinking at least one day a month.

Means, Standard Deviations, and Intercorrelations of Study Variables

The sequential logistic regression analysis revealed that having a previous suicide attempt was significantly associated with a greater likelihood of being a solitary binge drinker as opposed to only a social binge drinker (OR = 3.76, p = .006; see Table 2). Severity of suicidal ideation also was significantly associated with being a solitary binge drinker. In the third step of the analysis, suicide attempt history remained significantly associated with a greater likelihood of being a solitary binge drinker even when suicidal ideation was controlled for (OR = 2.76, p = .046), suggesting that both suicidality (suicidal ideation and attempts) variables were independently associated with a greater likelihood of being a solitary binge drinker.

Logistic Regression Predicting Being a Solitary Binge Drinker

Separate hierarchical multiple regression analyses were conducted to examine the association of suicidality with frequency of social drinking and with drinks per social drinking day. Frequency of drinking in social contexts was not significantly associated with a previous suicide attempt (ΔR² < .001, p = .85) or with severity of suicidal ideation (ΔR² = .005, p = .33). Similarly, drinks per social drinking day was not significantly associated with a previous suicide attempt (ΔR² = .001, p = .61) or with severity of suicidal ideation (ΔR² = .003, p = .47).

Given the higher suicidality among solitary binge drinkers, it was possible that social and solitary binge drinkers differed in the strength of association between social drinking context and suicidality. Therefore, a hierarchical regression analysis was conducted to examine potential differences between binge drinking groups in these associations. In the first step, age, gender, ethnicity, binge drinking group, previous suicide attempt, and suicidal ideation (mean centered) were entered into the model. In the second step, two cross-product interaction terms were entered: Binge Drinking Group × Suicide Attempt and Binge Drinking Group × Suicidal Ideation. The addition of these interaction terms did not add significantly to the regression models for frequency of social drinking (ΔR² = .014, p = .28) or drinks per social drinking day (ΔR² = .012, p = .30), indicating that there was not a significant difference between social and solitary binge drinkers. These analyses suggest that for both binge drinking groups, neither frequency nor amount of drinking in social contexts was significantly associated with suicidality.

The hierarchical multiple regression analyses examining the association of a previous suicide attempt and severity of suicidal ideation with solitary drinking frequency and drinks per solitary drinking day revealed that a previous suicide attempt was associated with more frequent solitary drinking (ΔR² = .041, p = .004; see Table 3) and with more drinks per solitary drinking day (ΔR² = .037, p = .008). Suicidal ideation was not significantly associated with frequency of solitary drinking once suicide attempt history was accounted for (ΔR² = .012, p = .12). However, suicidal ideation was associated with more drinks per solitary drinking day (ΔR² = .033, p = .010). Finally, repeating these analyses while controlling for data source (Study 1 vs. 2) revealed the same patterns of significance for the suicidality variables as reported above. The final step in the models containing the interaction terms of data source by suicide attempt history and suicidal ideation were nonsignificant in all analyses (ΔR² between .002 and .015), suggesting that results found do not differ significantly by data source despite differences in the selection criteria between studies.

Multiple Regression Analyses Predicting Solitary Drinking

The findings of this study suggest that binge drinking students with a suicide attempt history are significantly more likely to engage in solitary binge drinking. Students with a previous suicide attempt were nearly four times more likely to be solitary as opposed to only social binge drinkers. Those experiencing greater severity of recent suicidal ideation also were more likely to be solitary binge drinkers. Both a previous suicide attempt and recent suicidal ideation were independently associated with a greater likelihood of solitary binge drinking.

A history of a suicide attempt was associated with more frequent solitary drinking and having more drinks per solitary drinking day. In contrast, suicide attempt history and severity of suicidal ideation were not associated with frequency of social drinking or the amount of alcohol consumed on social drinking days for either social or solitary binge drinkers. The findings are consistent with that of a previous study that found that severity of suicidal ideation was significantly associated with solitary, but not social, binge drinking among underage students with a history of suicidal ideation (Gonzalez, Collins et al., 2009), and extends the findings to emerging adult students who were not selected for their suicidal ideation history.

Future studies are needed to examine how or why suicidal ideation and a history of a suicide attempt are associated with solitary binge drinking. One potential way that a previous suicide attempt and suicidal ideation may be related to solitary binge drinking is through drinking to cope with negative affect. According to motivational models of alcohol use, drinking to cope is motivated by efforts to escape, avoid, or lessen negative affect (Cooper, Frone, Russell, & Mudar, 1995). Drinking to cope appears to motivate binge drinking in the absence of the social influences commonly associated with drinking among emerging adult students (Christiansen et al., 2002). Consistent with this notion, frequency of solitary binge drinking in a previous study was found to be associated with severity of suicidal ideation and motivated by drinking to cope (Gonzalez, Collins et al., 2009).

Motivational models of alcohol use also suggest that coping-skill deficits contribute to a reliance on alcohol to cope with negative affect (Cooper et al., 1995; Cooper, Agocha, & Sheldon, 2000; Cox & Klinger, 1988). Similarly, cognitive–behavioral models of suicidality note the important role of coping-skill deficits in suicidal ideation and behavior (Reinecke, 2006; Rudd, 2006). Consistent with these models, suicidal ideation and behavior and drinking to cope are associated with greater use of avoidant coping strategies (Britton, 2004; Edwards & Holden, 2001; Reinecke, DuBois, & Schultz, 2001; Williams & Kleinfelter, 1989). Individuals with a history of a suicide attempt may be more likely to engage in solitary binge drinking in order to cope with distress, as well as be more likely to suffer from distress compared with social binge drinkers, in part owing to poorer coping skills. Future studies are needed to examine this possibility, as well as other potential links between solitary drinking and suicidality.

An important limitation of this study was the cross-sectional design. Because of this it is not known whether suicidal ideation motivates solitary binge drinking, solitary binge drinking plays a causal role in suicide attempts, or if these relationships are reciprocal or indirect. Another limitation was the overrepresentation of women in the sample. This study also was limited to emerging adult college binge drinkers and therefore may not generalize to noncollege student emerging adults or nonbinge drinking students. Studies are needed to examine solitary binge drinking in relation to negative affect and suicidality with noncollege population samples.

In conclusion, the current findings suggest that solitary binge drinkers are in particular need of suicide prevention efforts. Binge drinking alone among individuals with a suicide attempt history and/or greater severity of suicidal ideation is alarming in regard to risk for suicidal behavior, as intoxication can impede adaptive coping, increase aggression, and worsen mood (Hufford, 2001). Solitary binge drinkers are in need of treatment efforts aimed at reducing their alcohol misuse, given their greater frequency of social binge drinking as well as additional episodes of solitary binge drinking. This may serve to reduce the likelihood for alcohol dependence, as well as suicide risk within this population.

Barrios, L. C., Everett, S. A., Simon, T. R., & Brener, N. D. (2000). Suicide ideation among US college students: Associations with other injury risk behaviors. Journal of American College Health, 48, 229–233. doi:10.1080/07448480009599309

Beck, A. T., Steer, R. A., & Brown, G. K. (1996). Manual for the Beck Depression Inventory, 2nd ed.San Antonio, TX: The Psychological Corporation.

Beck, A. T., Steer, R. A., Kovacs, M., & Garrison, B. (1985). Hopelessness and eventual suicide: A 10-year prospective study of patients hospitalized with suicidal ideation. The American Journal of Psychiatry, 142, 559–563.

Borges, G., Angst, J., Nock, M. K., Ruscio, A. M., & Kessler, R. C. (2008). Risk factors for the incidence and persistence of suicide-related outcomes: A 10-year follow-up study using the National Comorbidity Surveys. Journal of Affective Disorders, 105, 25–33. doi:10.1016/j.jad.2007.01.036

Borges, G., Walters, E. E., & & Kessler, R. C. (2000). Associations of substance use, abuse and dependence with subsequent suicidal behavior. American Journal of Epidemiology, 151, 781–789.

Brady, J. (2006). The association between alcohol misuse and suicidal behaviour. Alcohol and Alcoholism, 41, 473–478. doi:10.1093/alcalc/agl060

Brener, N. D., Hassan, S. S., & Barrios, L. C. (1999). Suicidal ideation among college students in the United States. Journal of Consulting and Clinical Psychology, 67, 1004–1008. doi:10.1037/0022-006X.67.6.1004

Britton, P. C. (2004). The relation of coping strategies to alcohol consumption and alcohol-related consequences in a college sample. Addiction Research & Theory, 12, 103–114. doi:10.1080/16066350310001613062

Brown, G. K., Beck, A. T., Steer, R. A., & Grisham, J. R. (2000). Risk factors for suicide in psychiatric outpatients: A 20-year prospective study. Journal of Consulting and Clinical Psychology, 68, 371–377. doi:10.1037/0022-006X.68.3.371

Cherpitel, C. J., Borges, G. L. G., & Wilcox, H. C. (2004). Acute alcohol use and suicidal behavior: A review of the literature. Alcoholism: Clinical and Experimental Research, 28, 18S–28S. doi:10.1097/01.ALC.0000127411.61634.14

Christiansen, M., Vik, P. W., & Jarchow, A. (2002). College student heavy drinking in social contexts versus alone. Addictive Behaviors, 27, 393–404. doi:10.1016/S0306-4603(01)00180-0

Cooper, M. L., Agocha, V. B., & Sheldon, M. S. (2000). A motivational perspective on risky behaviors: The role of personality and affect regulatory processes. Journal of Personality, 68, 1059–1088. doi:10.1111/1467-6494.00126

Cooper, M. L., Frone, M. R., Russell, M., & Mudar, P. (1995). Drinking to regulate positive and negative emotions: A motivational model of alcohol use. Journal of Personality and Social Psychology, 69, 990–1005. doi:10.1037/0022-3514.69.5.990

Cotton, C. R., Peters, D. K., & Range, L. M. (1995). Psychometric properties of the Suicidal Behaviors Questionnaire. Death Studies, 19, 391–397. doi:10.1080/07481189508252740

Cox, W. M., & Klinger, E. (1988). A motivational model of alcohol use. Journal of Abnormal Psychology, 97, 168–180. doi:10.1037/0021-843X.97.2.168

Edwards, M. J., & Holden, R. R. (2001). Coping, meaning in life, and suicidal manifestations: Examining gender differences. Journal of Clinical Psychology, 57, 1517–1534. doi:10.1002/jclp.1114

Furr, S. R., Westefeld, J. S., McConnell, G. N., & Jenkins, J. M. (2001). Suicide and depression among college students: A decade later. Professional Psychology: Research and Practice, 32, 97–100. doi:10.1037/0735-7028.32.1.97

Gonzalez, V. M., Bradizza, C. M., & Collins, R. L. (2009). Drinking to cope as a statistical mediator in the relationship between suicidal ideation and alcohol outcomes among underage college drinkers. Psychology of Addictive Behaviors, 23, 443–451. doi:10.1037/a0015543

Gonzalez, V. M., Collins, R. L., & Bradizza, C. M. (2009). Solitary and social heavy drinking, suicidal ideation, and drinking motives in underage college drinkers. Addictive Behaviors, 34, 993–999. doi:10.1016/j.addbeh.2009.06.001

Gonzalez, V. M., Reynolds, B., & Skewes, M. C. (2011). Role of impulsivity in the relationship between depression and alcohol problems among emerging adult college drinkers. Experimental and Clinical Psychopharmacology, 19, 303–313. doi:10.1037/a0022720

Gonzalez, V. M., & Skewes, M. C. (2011). Solitary heavy drinking, social relationships, and negative mood regulation in college drinkers. Manuscript submitted for publication.

Gutierrez, P. M., Osman, A., Kopper, B. A., Barrios, F. X., & Bagge, C. L. (2000). Suicide risk assessment in a college student population. Journal of Counseling Psychology, 47, 403–413. doi:10.1037/0022-0167.47.4.403

Harris, E. C., & Barraclough, B. (1997). Suicide as an outcome for mental disorders. A meta-analysis. British Journal of Psychiatry, 170, 205–228. doi:10.1192/bjp.170.3.205

Hufford, M. R. (2001). Alcohol and suicidal behavior. Clinical Psychology Review, 21, 797–811. doi:10.1016/S0272-7358(00)00070-2

Jarvis, B. G. (2006). MediaLab(Version 2006.2.40) [Computer software]. New York, NY: Empirisoft.

Knight, J. R., Wechsler, H., Kuo, M., Seibring, M., Weitzman, E. R., & Schuckit, M. A. (2002). Alcohol abuse and dependence among U.S. college students. Journal of Studies on Alcohol, 63, 263–270.

Kuntsche, E., Knibbe, R., Gmel, G., & Engels, R. (2005). Why do young people drink? A review of drinking motives. Clinical Psychology Review, 25, 841–861. doi:10.1016/j.cpr.2005.06.002

Kuo, W. H., Gallo, J. J., & Eaton, W. W. (2004). Hopelessness, depression, substance disorder, and suicidality: A 13-year community-based study. Social Psychiatry and Psychiatric Epidemiology, 39, 497–501.

Kuo, W. H., Gallo, J. J., & Tien, A. Y. (2001). Incidence of suicide ideation and attempts in adults: The 13-year follow-up of a community sample in Baltimore, MD. Psychological Medicine: A Journal of Research in Psychiatry and the Allied Sciences, 31, 1181–1191. doi:10.1017/S0033291701004482

LaBrie, J. W., Hummer, J. F., & Pedersen, E. R. (2007). Reasons for drinking in the college student context: The differential role and risk of the social motivator. Journal of Studies on Alcohol and Drugs, 68, 393–398.

Levy, J. C., & Deykin, E. Y. (1989). Suicidality, depression, and substance abuse in adolescence. The American Journal of Psychiatry, 146, 1462–1467.

Mohr, C. D., Armeli, S., Tennen, H., Carney, M. A., Affleck, G., & Hromi, A. (2001). Daily interpersonal experiences, context, and alcohol consumption: Crying in your beer and toasting good times. Journal of Personality and Social Psychology, 80, 489–500. doi:10.1037/0022-3514.80.3.489

National Institute on Alcohol Abuse and AlcoholismTask Force on Recommended Questions of the National Council on Alcohol Abuse and Alcoholism. (2003, October). Recommended Sets of Alcohol Consumption Questions. Retrieved from http://www.niaaa.nih.gov.offcampus.lib.washington.edu/Resources/ResearchResources/TaskForce.htm

Oquendo, M. A., Galfalvy, H., Russo, S., Ellis, S. P., Grunebaum, M. F., Burke, A., & Mann, J. J. (2004). Prospective study of clinical predictors of suicidal acts after a major depressive episode in patients with major depressive disorder or bipolar disorder. The American Journal of Psychiatry, 161, 1433–1441. doi:10.1176/appi.ajp.161.8.1433

Osman, A., Bagge, C. L., Gutierrez, P. M., Konick, L. C., Kopper, B. A., & Barrios, F. X. (2001). The Suicidal Behaviors Questionnaire–revised (SBQ-R): Validation with clinical and nonclinical samples. Assessment, 8, 443–454. doi:10.1177/107319110100800409

Osman, A., Kopper, B. A., Linehan, M. M., Barrios, F. X., Gutierrez, P. M., & Bagge, C. L. (1999). Validation of the Adult Suicidal Ideation Questionnaire and the Reasons for Living Inventory in an adult psychiatric inpatient sample. Psychological Assessment, 11, 115–123. doi:10.1037/1040-3590.11.2.115

Reinecke, M. (2006). Problem solving: A conceptual approach to suicidality and psychotherapy. In T. E.Ellis (Ed.), Cognition and suicide: Theory, research, and therapy (pp. 237–260). Washington, DC: American Psychological Association. doi:10.1037/11377-011

Reinecke, M. A., DuBois, D. L., & Schultz, T. M. (2001). Social problem solving, mood, and suicidality among inpatient adolescents. Cognitive Therapy and Research, 25, 743–756. doi:10.1023/A:1012971423547

Reynolds, W. M. (1991a). Adult Suicidal Ideation Questionnaire: Professional manual. Odessa, FL: Psychological Assessment Resources.

Reynolds, W. M. (1991b). Psychometric characteristics of the Adult Suicidal Ideation Questionnaire in college students. Journal of Personality Assessment, 56, 289–307. doi:10.1207/s15327752jpa5602_9

Rudd, M. D. (2006). Fluid vulnerability theory: A cognitive approach to understanding the process of acute and chronic suicide risk. In T. E.Ellis (Ed.), Cognition and suicide: Theory, research, and therapy. (pp. 355–368). Washington, DC: American Psychological Association. doi:10.1037/11377-016

Slutske, W. S. (2005). Alcohol use disorders among US college students and their non-college-attending peers. Archives of General Psychiatry, 62, 321–327. doi:10.1001/archpsyc.62.3.321

Stephenson, H., Pena-Shaff, J., & Quirk, P. (2006). Predictors of college student suicidal ideation: Gender differences. College Student Journal, 40, 109–117.

Stewart, S. H., Zeitlin, S. B., & Samoluk, S. B. (1996). Examination of a three-dimensional drinking motives questionnaire in a young adult university student sample. Behaviour Research and Therapy, 34, 61–71. doi:10.1016/0005-7967(95)00036-W

Timberlake, D. S., Hopfer, C. J., Rhee, S. H., Friedman, N. P., Haberstick, B. C., Lessem, J. M., & Hewitt, J. K. (2007). College attendance and its effect on drinking behaviors in a longitudinal study of adolescents. Alcoholism: Clinical and Experimental Research, 31, 1020–1030. doi:10.1111/j.1530-0277.2007.00383.x

Wechsler, H., Davenport, A., Dowdall, G., & Moeykens, B. (1994). Health and behavioral consequences of binge drinking in college: A national survey of students at 140 campuses. JAMA: Journal of the American Medical Association, 272, 1672–1677. doi:10.1001/jama.1994.03520210056032

Wilcox, H. C., Conner, K. R., & Caine, E. D. (2004). Association of alcohol and drug use disorders and completed suicide: An empirical review of cohort studies. Drug and Alcohol Dependence, 76, S11–S19. doi:10.1016/j.drugalcdep.2004.08.003

Williams, J. G., & Kleinfelter, K. J. (1989). Perceived problem-solving skills and drinking patterns among college students. Psychological Reports, 65, 1235–1244. doi:10.2466/pr0.1989.65.3f.1235

Submitted: April 7, 2011 Revised: December 7, 2011 Accepted: December 8, 2011

Correction to Gonzalez (2012)

In the article “Association of Solitary Binge Drinking and Suicidal Behavior Among Emerging Adult College Students,” by Vivian M. Gonzalez (Psychology of Addictive Behaviors, Advance online publication, January 30, 2012. doi: 10.1037/a0026916), there is an error in the introductory paragraph. The number of students who had seriously considered attempting suicide in the Barrios, Everett, Simon, & Brener (2000) study should have been reported as 11.4%, not 1.4%. Additionally, in the Participants section, data for the study were collected from March 2009 to September 2010, not March 2009 to January 2010 as reported.

Associations of Descriptive and Reflective Injunctive Norms With Risky College Drinking / BRIEF REPORT

Acknowledgement: Supported by a Small Grant from the Alcohol and Drug Abuse Institute (Grant No. 65-1951), a National Institute on Alcohol Abuse and Alcoholism (NIAAA) Career Transition Award (Grant No. K22 AA018384), and the NIAAA (Grant No. R01 AA012547).

The negative consequences associated with the consistently high rates of heavy episodic drinking (i.e., ≥ 4 drinks for women and ≥ 5 drinks for men; Dawson, Grant, Stinson, & Chou, 2004; Nelson, Xuan, Lee, Weitzman, & Wechsler, 2009) among college students are well-documented. Heavy episodic drinking affects both college drinkers (e.g., accidents and falls resulting in injury; risky sex; unwanted sexual advances, rape, and sexual assault; driving while intoxicated; problems with authorities; Abbey, 2002; Hingson, Heeren, Zakocs, Kopstein, & Wechsler, 2002) and their communities (e.g., sleep disruption; property damage; verbal, physical, or sexual violence; Langley, Kypri, & Stephenson, 2003; Wechsler & Nelson, 2008). The pressing severity of the problem has led researchers to explore correlates and predictors of heavy-episodic drinking among college students—including family history, demographics, personality factors, drinking motives, expectancies, attitudes, peer influences, drinking onset, and, perhaps most prolifically, perceived norms—to better understand how to effectively intervene (Baer, 2002; Borsari, Murphy, & Barnett, 2007).

Dozens of studies have shown strong associations between norms and college drinking (see Borsari & Carey, 2003; Borsari & Carey, 2001, for reviews). Most of these studies have evaluated the importance of descriptive norms, or the target student’s beliefs about how much other reference groups drink, and have collectively shown that college students consistently overestimate referents’ drinking (Borsari & Carey, 2003). Descriptive norms have been shown to be positively associated with alcohol outcomes (e.g., Baer, Stacy, & Larimer, 1991; Larimer et al., 2009; Larimer et al., 2011; Larimer, Turner, Mallett, & Geisner, 2004; Perkins, 2007; Perkins, Haines, & Rice, 2005).

Injunctive norms, or how much the target student believes referents approve or disapprove of drinking in general, have been researched to a lesser extent, but studies have shown similar outcomes. Specifically, college students perceive reference groups to approve of drinking more than is actually the case (Borsari & Carey, 2003; DeMartini, Carey, Lao, & Luciano, 2011; LaBrie, Napper, & Ghaidarov, 2012). On the other hand, the association between injunctive norms and drinking outcomes is more complex. Studies have shown that self-other discrepancies (i.e., the perceived difference between the target’s and the referents’ behaviors and attitudes) created by injunctive norms are larger than for descriptive norms (Borsari & Carey, 2003). The association of injunctive norms with drinking outcomes varies by reference group: the closer the referents, the greater the positive association between perceived referent approval of drinking and the target student’s drinking and alcohol-related problems (LaBrie, Hummer, Neighbors, & Larimer, 2010; Neighbors et al., 2008).

Reflective injunctive norms are a more personally referenced version of injunctive norms, and they were originally introduced as “subjective norms” in the theory of reasoned action (Ajzen & Fishbein, 1980) and, later, in the theory of planned behavior (Ajzen, 1991).

Reflective injunctive norms represent the extent to which the target believes referents approve of the target’s own drinking (Ajzen, 2002). As applied in the theory of planned behavior, reflective injunctive norms have been shown to evince a positive association with intent to engage in risky drinking, and, in some studies, an indirect effect on risky drinking via intent (Collins & Carey, 2007; Collins, Witkiewitz, & Larimer, 2011; Huchting, Lac, & LaBrie, 2008; Johnston & White, 2003).

The aim of this study was to assess the relative predictive power of parallel descriptive and reflective injunctive norms across various peer groups and thereby to gain further insight into which types of norms and reference groups are most highly concurrently correlated with risky drinking. It was hypothesized that both reflective injunctive and descriptive norms would be significantly positively correlated with risky drinking outcomes (i.e., peak drinking quantity, heavy drinking episode [HDE] frequency, and alcohol-related problems), and that more proximal reference group norms would be more highly predictive than more distal reference group norms.

Participants

Participants were 837 (63.9% female, 0.1% transgender) college students at two, 4-year universities on the U.S. west coast who participated in a longitudinal parent study (response rate = 70%; Collins et al., 2011). Please see Table 1 for sample description.

Sociodemographic Sample Description (N = 837)

Measures

The Personal Information Questionnaire was used to assess participants’ age, gender, year in college, race and ethnicity, and membership in an on-campus Greek organization. This measure was used to describe the baseline sample.

The Reflective Injunctive Norms Questionnaire was made up of indicators from the Subjective Norms Questionnaire, a measure based on suggestions by Ajzen (2002), and modified from a previous study (Collins & Carey, 2007). Participants rated how much various referents would approve or disapprove of their “drinking until you get drunk” in the next 30 days, on a scale ranging from 1 (highly disapprove) to 5 (highly approve). Target groups included “an average college student at your university,” “your closest friend,” “your closest family member,” and “the person whose opinion you value most.” The question stem was rephrased to assess descriptive norms for each of the reference groups (e.g., “Will an average college student at your university drink until they get drunk at least once in the next 30 days?”). Participants could agree or disagree with these statements (1 = strongly disagree to 5 = strongly agree). Items for descriptive (α = .96) and reflective injunctive (α = .96) norms showed good internal consistency.

The Frequency−Quantity Questionnaire (adapted from Borsari & Carey, 2000; Collins, Carey, & Sliwinski, 2002; Dimeff, Baer, Kivlahan, & Marlatt, 1999) includes open-ended items assessing participants’ self-aggregated alcohol frequency and quantity of alcohol consumption in the past 30 days.

The Timeline Followback Questionnaire (TLFB; Sobell & Sobell, 1992) was used to aggregate HDE frequency in the past 30 days. The TLFB is a set of monthly calendars that allows for a retrospective evaluation of drinking for each day of the previous month.

Alcohol-related problems were assessed using the Rutgers Alcohol Problems Index (White & Labouvie, 1989), which asks participants how often (0 = never to 4 = more than 10 times) they experienced 23 drinking-related consequences over the previous 30 days. The overall score showed acceptable internal consistency (α = .94).

Procedure

Potential participants for the current study were emailed invitations. Invitations included the study URL and a randomly generated personal identification number with which participants logged into the secure study website. Participants provided informed consent and completed the baseline assessment, which included the measures noted above and took approximately 45 min. They were paid $20 for their time.

Data Analysis Plan

Three regression models were conducted in Stata, Version 11.2 to determine the relative contributions of reflective injunctive and descriptive norms in the prediction of risky drinking outcomes (i.e., alcohol quantity during the heaviest drinking occasion in the past 30 days [peak alcohol quantity], HDE, and alcohol-related problems). Because drinking outcome variables were overdispersed, positively skewed count/integer responses and evinced a preponderance of zeros, zero-inflated negative binomial (ZINB) models were used (Cameron & Trivedi, 1998). ZINB is a subset of generalized linear models for count outcomes that are positively skewed and have more zero responses than would be expected given the distribution. ZINB models two processes: the zero-inflated portion of the model, which is a Bernoulli trial to determine the probability that an observation is consistently zero, and a negative binomial portion of the model, which determines the association if the observation is a feasible count response predicted by the negative binomial distribution (Hardin & Hilbe, 2007).

Descriptive statistics for the predictors and outcomes are presented in Table 2, and zero-order correlations are listed in Table 3.

Descriptive Statistics for Predictors and Outcomes

Zero-Order Correlations (Bivariate Spearman’s ρ) Between Norms and Risky Drinking

ZINB Model Outcomes

Peak alcohol quantity

The omnibus model was significant, χ²(8, n = 831) = 87.51, p < .001, Vuong z = 7.68, p < .001, Nagelkerke pseudo R² = .30. The negative binomial portion of the model indicated that descriptive norms for a typical college student inversely predicted peak alcohol quantity, whereas descriptive norms for a respondent’s closest friend positively predicted peak alcohol quantity. Reflective injunctive norms for one’s closest friend were positively associated with peak alcohol quantity (see Table 4 for model parameters). The zero-inflated portion of the model indicated that both descriptive and reflective injunctive norms for one’s closest friend inversely predicted zero inflation. Therefore, the greater the agreement that one’s closest friend would drink until drunk and approve of the respondent drinking until drunk, the lower the likelihood the student belonged to the “consistent zero,” or abstinent part of the distribution. Additionally, the greater the perceived approval by the person whose opinion they valued the most, the less likely respondents were to belong to the consistent zero group (see Table 4).

Zero-Inflated Negative Binomial Model ParametersZero-Inflated Negative Binomial Model Parameters

Heavy drinking episodes

The omnibus model was significant, χ²(8, n = 831) = 53.57, p < .001, Vuong z = 4.42, p < .001, Nagelkerke pseudo R² = .30. The negative binomial portion of the model indicated that descriptive norms for a typical college student inversely predicted HDE, whereas descriptive norms for a respondent’s closest friend positively predicted HDE. Regarding reflective injunctive norms, similar relationships existed for norms regarding typical students and one’s closest friend (see Table 4). The zero-inflated portion of the model indicated that both descriptive and reflective injunctive norms for one’s closest friend inversely predicted zero inflation (see Table 4).

Alcohol-related problems

The omnibus model was significant, χ²(8, n = 797) = 16.45, p = .04, Vuong z = 4.01, p < .001, Nagelkerke pseudo R² = .20. None of the negative binomial results was significant for predicting alcohol-related problems (see Table 4). The zero-inflated portion of the model indicated that descriptive norms for one’s closest friend and the person whose opinion the respondent valued most inversely predicted zero inflation (see Table 4).

The aims of this study were to test the relative predictive power of parallel descriptive and reflective injunctive norms across various peer groups. It was hypothesized that both reflective injunctive and descriptive norms would be associated with risky drinking outcomes (i.e., peak drinking quantity, HDE frequency, and alcohol-related problems), and that more proximal referents would be more highly predictive than more distal referents.

Summary of Current Findings

Findings largely corresponded to hypotheses. Zero-order correlations showed that both descriptive and reflective injunctive norms regarding closer referents (in descending level of association: closest friend, person whose opinion you value most, and closest family member) showed stronger positive correlations with risky drinking and related problems than norms regarding typical college students.

Next, our multivariate analyses afforded the unique opportunity to differentiate between two aspects of risky drinking: the odds of typically engaging in nonrisky versus risky drinking as well as the extent of risky drinking. Regarding the former, we were able to note which types of norms predicted drinking outcomes for people who were likely never to drink, experience HDE, or alcohol-related problems. Specifically, the norm for one’s closest friend was the primary predictor among descriptive norms, such that greater belief one’s closest friend would drink until they got drunk was associated with lower likelihood of being either abstinent or a lighter drinker. Greater conviction that both one’s closest friend and person whose opinion one values most would drink until drunk was also associated with a lower likelihood of nonproblem drinking. Regarding reflective injunctive norms, we found that greater agreement that one’s closest friend or person whose opinion one values most would approve of one drinking until drunk was associated with lower likelihood of being either abstinent or a light drinker.

Second, we examined the prediction of the extent of risky drinking by descriptive and injunctive norms. We found that descriptive and reflective injunctive norms for one’s closest friend served as the most consistent, positive predictors of risky drinking outcomes, with reflective injunctive norms for the person whose opinion respondents value most being a similarly consistent positive predictor. Unlike in the zero-order correlations, however, closest family member was never significantly associated with risky drinking outcomes, and neither descriptive nor reflective injunctive norm was associated with the extent of one’s experience of alcohol-related problems. Finally, descriptive norms regarding a typical college student were inversely—not positively—associated with risky drinking outcomes in the multivariate models.

Findings in the Context of the Norms Literature

Our findings replicated and extended those of other recent studies of norms and their relative weight across various reference groups (Larimer et al., 2009; Larimer et al., 2011) and norm types (LaBrie et al., 2010; Neighbors et al., 2007; Neighbors et al., 2008). Regarding the similarities between other studies and the current study, we noted that both descriptive and reflective injunctive norms using typical college students as the reference group were positively associated with risky drinking in zero-order correlations—although not as strongly positively correlated as closer reference groups. When examined in the context of other reflective injunctive and descriptive norms, however, norms for typical students were inversely associated with experience of HDE and peak alcohol quantity. A similar pattern of findings has been found across a few studies whose authors have attributed it to a potential negative suppressor effect (cf. Neighbors et al., 2007; Neighbors et al., 2008). Although more studies are needed to further parse and confirm this finding, it may be concluded that norms regarding closer referents and not typical college students are more powerful and consistent predictors of risky drinking outcomes.

Although it replicated findings from previous studies, the current study also expanded on these findings to include new reference groups (e.g., “person whose opinion you value most,” “closest family member”), a target behavior that is more specific to HDE experienced by college students (i.e., “drinking until you got drunk”), and a different type of injunctive norm (i.e., reflective injunctive norms) also known as “subjective norms” in the theory of planned behavior (Ajzen, 1991). Although reflective norms have been explored in a couple of prior studies in the college drinking norms literature, this has only been done in the context of opposite-sex perceptions and has not been personally referenced (Hummer, LaBrie, Lac, Sessoms, & Cail, 2012; LaBrie, Cail, Hummer, Lac, & Neighbors, 2009). Because they capture people’s perceptions of others’ approval of their own behavior, the current findings suggest reflective injunctive norms may be personally relevant and affectively salient.

Study Limitations

Because we did not additionally assess injunctive norms, it is impossible to compare the relative contributions of the reference groups’ approval of drinking until drunk in general (injunctive norms) versus approval of the target student’s drinking until drunk (reflective injunctive norms). Future studies may include both types of injunctive norms to understand their relative predictive abilities.

The norms questionnaires did not ask participants to indicate exactly who certain referents were (e.g., whom participants had in mind when considering their “closest family member” or “person whose opinion they valued most”). Nonetheless, we were able to address the research question: what types of norms are most predictive of risky drinking outcomes. Future studies may incorporate qualitative methods to describe these most salient referents in greater detail.

Finally, this study is a cross-sectional representation of the associations between college drinking and norms. To better understand how norms may affect college drinking trajectories, it will be important to replicate such findings in longitudinal studies. Despite these limitations, however, the current study was able to introduce new reference groups, a new target behavior, and a different type of injunctive norm. In doing so, we were able to replicate and expand on the existing college norms literature.

Conclusion and Future Directions

This study indicates that descriptive and reflective injunctive norms are most highly, positively correlated with college drinking when they involve referents who are closer to the target college drinkers (i.e., closest friend and person whose opinion you value the most). These findings may contribute to the knowledge base for enhancing normative reeducation and personalized normative feedback interventions to include more personally salient and powerful normative information.

Future studies should include assessment of both injunctive and reflective injunctive norms to better characterize the difference between these two aspects of social approval, and to understand whether one may be more affectively salient than the other. In turn, future studies of personalized normative feedback may include feedback on closer referents to create more powerful interventions that inspire greater and potentially more lasting drinking behavior change for college students. This suggested direction is challenging, because it is much more difficult to gather information on specific referents to assess the normative discrepancy. On the other hand, future studies could explore feedback on reflective injunctive norms without relying on normative discrepancy. It is possible that simply making college drinkers aware of their normative beliefs could build discrepancy (e.g., self-other or ideal current drinking) that could, in turn, influence drinking behavior.

Abbey, A. (2002). Alcohol-related sexual assault: A common problem among college students. Journal of Studies on Alcohol, Supplement, 14, 118–128.

Ajzen, I. (1991). The theory of planned behavior. Organizational Behavior and Human Decision Processes, 50, 179–211. doi:10.1016/0749-5978(91)90020-T

Ajzen, I. (2002). Constructing a theory of planned behavior questionnaire. Retrieved from http://www.people.umass.edu/aizen/pdf/tpb.measurement.pdf

Ajzen, I., & Fishbein, M. (1980). Understanding attitudes and predicting social behavior. Englewood Cliffs, NJ: Prentice Hall.

Baer, J. S. (2002). Student factors: Understanding individual variation in college drinking. Journal of Studies on Alcohol, Supplement, 14, 40–53.

Baer, J. S., Stacy, A., & Larimer, M. (1991). Biases in the perception of drinking norms among college students. Journal of Studies on Alcohol, 52, 580–586.

Borsari, B., & Carey, K. (2003). Descriptive and injunctive norms in college drinking: A meta-analytic integration. Journal of Studies on Alcohol, 64, 331–341.

Borsari, B., & Carey, K. B. (2000). Effects of a brief motivational intervention with college student drinkers. Journal of Consulting and Clinical Psychology, 68, 728–733. doi:10.1037/0022-006X.68.4.728

Borsari, B., & Carey, K. B. (2001). Peer influences on college drinking: A review of the research. Journal of Substance Abuse, 13, 391–424. doi:10.1016/S0899-3289(01)00098-0

Borsari, B., Murphy, J. G., & Barnett, N. P. (2007). Predictors of alcohol use during the first year of college: Implications for prevention. Addictive Behaviors, 32, 2062–2086. doi:10.1016/j.addbeh.2007.01.017

Cameron, A. C., & Trivedi, P. K. (1998). Regression analysis of count data. Cambridge, England: Cambridge University Press. doi:10.1017/CBO9780511814365

Collins, S. E., & Carey, K. B. (2007). The theory of planned behavior as a model of heavy episodic drinking among college students. Psychology of Addictive Behaviors, 21, 498–507. doi:10.1037/0893-164X.21.4.498

Collins, S. E., Carey, K. B., & Sliwinski, M. J. (2002). Mailed personalized normative feedback as a brief intervention for at-risk college drinkers. Journal of Studies on Alcohol, 63, 559–567.

Collins, S. E., Witkiewitz, K., & Larimer, M. E. (2011). The theory of planned behavior as a predictor of growth in risky college drinking. Journal of Studies on Alcohol and Drugs, 72(2), 322–332.

Dawson, D. A., Grant, B. F., Stinson, F. S., & Chou, P. S. (2004). Another look at heavy episodic drinking and alcohol use disorders among college and noncollege youth. Journal of Studies on Alcohol, 65, 477–488.

DeMartini, K. S., Carey, K. B., Lao, K., & Luciano, M. (2011). Injunctive norms for alcohol-related consequences and protective behavioral strategies: Effects of gender and year in school. Addictive Behaviors, 36, 347–353. doi:10.1016/j.addbeh.2010.12.009

Dimeff, L. A., Baer, J. S., Kivlahan, D. R., & Marlatt, G. A. (1999). Brief Alcohol Screening and Intervention for College Students (BASICS): A harm reduction approach. New York, NY: Guilford Press.

Hardin, J. W., & Hilbe, J. M. (2007). Generalized linear models and extensions (2nd ed.). College Station, TX: Stata Press.

Hingson, R. W., Heeren, T., Zakocs, R. C., Kopstein, A., & Wechsler, H. (2002). Magnitude of alcohol-related mortality and morbidity among U.S. college students ages 18–24. Journal of Studies on Alcohol, 63, 136–144.

Huchting, K., Lac, A., & LaBrie, J. W. (2008). An application of the theory of planned behavior to sorority alcohol consumption. Addictive Behaviors, 33, 538–551. doi:10.1016/j.addbeh.2007.11.002

Hummer, J. F., LaBrie, J. W., Lac, A., Sessoms, A., & Cail, J. (2012). Estimates and influences of reflective opposite-sex norms on alcohol use among a high-risk sample of college students: Exploring Greek-affiliation and gender effects. Addictive Behaviors, 37, 596–604. doi:10.1016/j.addbeh.2011.11.027

Jaccard, J., & Wan, C. K. (1996). LISREL approaches to interaction effects in multiple regression. Thousand Oaks, CA: Sage.

Johnston, K. L., & White, K. M. (2003). Binge-drinking: A test of the role of group norms in the theory of planned behaviour. Psychology and Health, 18, 63–77. doi:10.1080/0887044021000037835

LaBrie, J. W., Cail, J., Hummer, J. F., Lac, A., & Neighbors, C. (2009). What men want: The role of reflective opposite-sex normative preferences in alcohol use among college women. Psychology of Addictive Behaviors, 23, 157–162. doi:10.1037/a0013993

LaBrie, J. W., Hummer, J. F., Neighbors, C., & Larimer, M. E. (2010). Whose opinion matters? The relationship between injunctive norms and alcohol consequences in college students. Addictive Behaviors, 35, 343–349. doi:10.1016/j.addbeh.2009.12.003

LaBrie, J. W., Napper, L. E., & Ghaidarov, T. M. (2012). Predicting driving after drinking over time among college students: The emerging role of injunctive normative perceptions. Journal of Studies on Alcohol and Drugs, 73, 726–730.

Langley, J. D., Kypri, K., & Stephenson, S. C. R. (2003). Secondhand effects of alcohol use among university students: Computerised survey. BMJ: British Medical Journal, 327, 1023–1024. doi:10.1136/bmj.327.7422.1023

Larimer, M. E., Kaysen, D. L., Lee, C. M., Kilmer, J. R., Lewis, M. A., Dillworth, T. M., . . .Neighbors, C. (2009). Evaluating level of specificity of normative referents in relation to personal drinking behavior. Journal of Studies on Alcohol and Drugs, Supplement, 16, 115–121.

Larimer, M. E., Neighbors, C., LaBrie, J. W., Atkins, D. C., Lewis, M. A., Lee, C. M., . . .Walter, T. (2011). Descriptive drinking norms: For whom does reference group matter?Journal of Studies on Alcohol and Drugs, 72, 833–843.

Larimer, M. E., Turner, A. P., Mallett, K. A., & Geisner, I. M. (2004). Predicting drinking behavior and alcohol-related problems among fraternity and sorority members: Examining the role of descriptive and injunctive norms. Psychology of Addictive Behaviors, 18, 203–212. doi:10.1037/0893-164X.18.3.203

Neighbors, C., Lostutter, T. W., Whiteside, U., Fossos, N., Walker, D. D., & Larimer, M. E. (2007). Injunctive norms and problem gambling among college students. Journal of Gambling Studies, 23, 259–273. doi:10.1007/s10899-007-9059-3

Neighbors, C., O’Connor, R. M., Lewis, M. A., Chawla, N., Lee, C. M., & Fossos, N. (2008). The relative impact of injunctive norms on college student drinking: The role of reference group. Psychology of Addictive Behaviors, 22, 576–581. doi:10.1037/a0013043

Nelson, T. F., Xuan, Z., Lee, H., Weitzman, E. R., & Wechsler, H. (2009). Persistence of heavy drinking and ensuing consequences at heavy drinking colleges. Journal of Studies on Alcohol and Drugs, 70, 726–734.

Perkins, H. W. (2007). Misperceptions of peer drinking norms in Canada: Another look at the “reign of error” and its consequences among college students. Addictive Behaviors, 32, 2645–2656. doi:10.1016/j.addbeh.2007.07.007

Perkins, H. W., Haines, M. P., & Rice, R. (2005). Misperceiving the college drinking norm and related problems: A nationwide study of exposure to prevention information, perceived norms and student alcohol misuse. Journal of Studies on Alcohol, 66, 470–478.

Sobell, L. C., & Sobell, M. B. (1992). Timeline followback: A technique for assessing self-reported ethanol consumption. In J.Allen & R. Z.Litten (Eds.), Measuring alcohol consumption: Psychosocial and biological methods (pp. 41–72). Totowa, NJ: Humana Press. doi:10.1007/978-1-4612-0357-5_3

Wechsler, H., & Nelson, T. F. (2008). What we have learned from the Harvard School of Public Health College Alcohol Study: Focusing attention on college student alcohol consumption and the environmental conditions that promote it. Journal of Studies on Alcohol and Drugs, 69, 481–490.

White, H. R., & Labouvie, E. W. (1989). Towards the assessment of adolescent problem drinking. Journal of Studies on Alcohol, 50, 30–37.

Submitted: September 14, 2012 Revised: March 11, 2013 Accepted: March 20, 2013

Baseline Reaction Time Predicts 12-Month Smoking Cessation Outcome in Formerly Depressed Smokers

Acknowledgement: This study was partially supported by National Institute on Drug Abuse Grant DA08511 to Richard A. Brown. We gratefully acknowledge Michelle Ricci for her able assistance in software programming. Portions of this article were presented at the fifth annual meeting of the Society for Research on Nicotine and Tobacco, San Diego, California, March, 1999.

The pathways to becoming a smoker are, no doubt, complex (e.g., Jamner et al., 2003; Kassel, Weinstein, Skitch, Veilleux, & Mermelstein, 2005). However, of the numerous factors believed to heighten vulnerability to smoking initiation, development of nicotine dependence, and smoking relapse, the role played by various forms of psychopathology and emotional distress appears particularly critical (Kassel, Stroud, & Paronis, 2003). That is, numerous studies have reliably found high smoking rates among selected populations of individuals with mental illness. For example, drawing upon a large, nationally representative sample in the United States, Lasser et al. (2000) found that individuals with a lifetime history of any psychiatric disorder had higher rates of lifetime and current smoking compared to individuals who had never suffered from mental illness. Indeed, other investigations have reported similar findings, demonstrating strong and reliable associations between psychiatric disorders—particularly depression (Kassel & Hankin, 2006)—and cigarette smoking among adults (Breslau, Kilbey, & Andreski, 1991; Covey, Glassman, & Stetner, 1998; Degenhardt & Hall, 2001).

Unfortunately, relapse is the modal outcome among those attempting to quit smoking (Piasecki, 2006). Whereas the best available treatments yield 1-year abstinence rates approaching 30–35%, even among smokers who successfully quit for a full year, as many as 40% eventually return to regular smoking (U.S. Department of Health and Human Services, 1990). Smokers who attempt to quit on their own fare even less well, with relapse rates ranging from 90% to 97% (Cohen et al., 1989; Hughes et al., 1992). Negative affect and symptoms of depression appear to be strongly implicated in these high relapse rates.

Identification of the extent to which depression predisposes to smoking relapse has emerged as a critical area of research inquiry in recent years. Indeed, a burgeoning research literature points to both strong between- and within-person associations between depressive symptoms and smoking relapse (see Kassel & Hankin, 2006; Kassel et al., 2003). The presence of clinically significant levels of negative affect and depressive symptoms is frequently predictive of relapse (Glassman et al., 1990; Hall, Munoz, Reus, & Sees, 1993). For instance, one study reported that the likelihood of quitting smoking was 40% lower among depressed smokers compared with nondepressed smokers (Anda et al., 1990). Glassman et al. (1990) found a quit rate of 14% for those meeting criteria for major depression, whereas 31% of participants with no psychiatric diagnosis successfully quit. Moreover, even in the absence of current symptomatology, history of depression may serve to heighten risk for both relapse (Covey, 1999; Glassman et al., 1993; though see Hitsman, Borrelli, McChargue, Spring, & Niaura, 2003) and recurrence of depressive symptomatology subsequent to cessation (Covey, Glassman, & Stetner, 1997). Indeed, history of recurrent major depression (i.e., having two or more past episodes) appears to be a particularly robust predictor of poor smoking cessation outcome (Brown et al., 2001; (Haas, Munoz, Humfleet, Reus, & Hall, 2004). Niaura et al. (2001) demonstrated that even low (subclinical) levels of depressive symptoms assessed at baseline among smokers enrolled in a cessation program were predictive of time to first cigarette smoked after attempted quitting.

One potential limitation of the rather extensive database examining the link between depression and smoking relapse is that depression (be it at the clinical or subclinical level) has been primarily assessed via clinical interview or self-report. There are clearly problems inherent in sole reliance on these types of measurement strategies. Although self-report measures are simple, inexpensive, and easy to use, they are susceptible to biases, social expectations, and attributions, and can ultimately measure only one's conscious experience of depressive symptomology. Correspondingly, a large literature demonstrates that depression—even depressive symptoms that fall short of clinical diagnosis—is reliably associated with a host of neurocognitive deficits (e.g., Porter, 2003; White, Myerson, & Hale, 1997). For example, Pardo, Pardo, Humes, and Posner (2006) recently conducted a study assessing depressed and nondepressed individuals' performance on tasks measuring phasic alerting and covert orientation of visuospatial attention. Their findings suggest that global slowing is a major cognitive deficit in depression. In fact, such findings support the results of a meta-analysis demonstrating that task-independent, generalized global slowing accounts for the apparent specific cognitive deficits found in unipolar depression (White et al., 1997).

Global slowing may be best viewed as a manifest difference in speeded trials between two comparison groups (Pardo et al., 2006). That is, global slowing represents a very real deficit in processing speed among individuals, revealed by comparison with an appropriate reference group. However, one should not conceive of global slowing as a phenomenon necessarily related to specific disorders (e.g., depression) or groups of people (e.g., older adults). Rather, global slowing has been observed across diverse samples, including sleep-deprived individuals, those who abuse central nervous system depressants (e.g., barbiturates and alcohol), and individuals with other psychiatric disorders, such as schizophrenia (Ryan, Russo, & Greeley, 1996).

Whereas psychomotor and cognitive slowing of the kind reported above are widely recognized as critical components of the depression syndrome, rarely are such cognitive measures utilized in the context of smoking cessation research. Because there is also reason to believe that, in recent years, the remaining population of smokers is hardening (i.e., becoming more nicotine dependent or exhibiting more comorbid psychopathology, including depression; Warner & Burns, 2003), it is incumbent upon tobacco researchers and clinicians alike to further elucidate the admittedly complex association between depressive symptomatology and smoking cessation. Toward this end, we prospectively examined the relationship between reaction time—an index of processing speed—and smoking cessation outcome in a group of formerly depressed smokers. Specifically, we were interested in ascertaining whether a simple measure of reaction time would correlate with a self-report measure of depression and add to the predictive utility of a self-report measure of depression in predicting long-term smoking cessation outcome.

Participants

Participants in the present study were recruited from the community and were already participating in a large treatment outcome study (N = 179) comparing standard smoking cessation treatment with treatment incorporating cognitive–behavioral therapy for depression (see Brown et al., 2001 for details). For the present study, 28 randomly selected participants agreed to take part in an investigation examining “cognitive factors related to smoking cessation success.” These smokers (mean age = 48.8 years, 58% female) had smoked an average of 29 years (SD = 10.4), smoked a mean of 28 cigarettes a day (SD = 14.3), had a mean Fagerstrom Test for Nicotine Dependence (FTND; Heatherton, Kozlowski, Frecker, & Fagerstrom, 1991) score of 7.4 (SD = 1.6), reported an average of 7.7 previous smoking cessation attempts (SD = 11.1), and presented with a mean Beck Depression Inventory (BDI; Beck, Ward, Mendelson, Mock, & Erbaugh, 1961) score of 8.7 (SD = 5.7). As assessed with the Structured Clinical Interview for DSM-III-R-Non-Patient Edition (SCID-NP; Spitzer, Williams, Gibbon, & First, 1990), all participants met diagnostic criteria for past, but not present, history of major depression.

Procedures

Participants were administered an emotional Stroop task (see Williams, Mathews, & MacLeod, 1996) prior to quitting smoking (T1) and immediately after their target quit date (T2). The emotional Stroop task assesses verbal response latency to colored words presented on a computer screen. Words either were of neutral semantic content (e.g., pencil, table, and carpet) or were emotionally laden target words (e.g., depression-related words, such as sad, dejected, and worried; anxiety-related words, such as nervous, worried, and anxious; anger-related words, such as angry, mad, and hostile; and smoking-related words, such as cigarette, smoke, and ashtray). For the purposes of the present analyses, we considered only subjects' response times (RTs; time to name out loud the color of presented words) to neutral, nonemotionally laden words at T1. As such, the assessed RTs served as a rather pure index of processing speed, unencumbered by potential emotional interference effects or by nicotine withdrawal (evident at T2).

After brief instruction and a practice trial conducted during the T1 visit, participants were presented a total of 120 (20 neutral) interspersed words, displayed one at a time on a video graphics array monitor in red, green, yellow, or blue. Their task was to indicate the color in which respective words appeared by saying the color of the word (thereby ignoring the semantic content of the stimulus) out loud as quickly as they could. Response latency was recorded via a microphone linked to a Macintosh software program designed to detect and record verbal latency to the nearest millisecond.

Thus, our primary dependent variable was 7-day point-prevalence abstinence rate (Hughes et al., 2003) assessed at 12 months posttreatment. The intent-to-treat principle was used, as was biochemical verification (expired air carbon monoxide and salivary cotinine), in order to confirm 12-month smoking status. Hence, participants were categorized as either abstinent or nonabstinent at 12 months posttreatment. Predictor variables included RT to name the color of only the affectively neutral words, nicotine dependence (FTND), and baseline (T1) depressive symptoms (BDI).

Even though we are interested in only the reaction time to the emotionally neutral words for the purposes of the present article, to place the RT measure in broader context, Figure 1 displays reaction times to all word categories at both T1 and T2. Initial correlational analyses are in Table 1. As anticipated, RT was significantly and positively correlated with depressive symptomatology at Visit 1 (p < .01); slower response times were associated with higher BDI scores. Moreover, RT was significantly correlated with 12-month point-prevalence abstinence rates (p < .01), such that slower response times to the naming of the color of neutral words were associated with a greater likelihood of nonabstinent smoking status. To explore the possibility that the observed relationship between RT and smoking outcome was simply attributable to the significant correlation between depressive symptoms and RT, we conducted a partial correlational analysis, in which we assessed the relationship between RT and 12-month outcome, while controlling for T1 BDI scores. The correlation remained significantly negative (pr = −.41, p < .04).

Figure 1. Reaction time displayed by visit (Visit 1 = prequit, Visit 2 = postquit) and word category. Error bars represent standard error of the mean. Dep = depression; Anx = anxiety; Smok = smoking.

Correlation Matrix of Variables

Depressive symptoms at both T1 and the 12-month follow-up were modestly, though nonsignificantly, associated with smoking outcome in the anticipated direction (both ps < .14); higher depressive symptoms were associated with smoking outcomes of nonabstinence. Of note, nicotine dependence (FTND) was not significantly correlated with 12-month smoking status (p = .11).

Outcome data revealed that of the 28 participants, 18 were nonabstinent at 12 months posttreatment, yielding a success rate of 36%. Next, we conducted an independent sample t test, in order to see whether 12-month smoking status was influenced by RTs assessed at Visit 1. Consistent with the correlational analysis, results revealed that those who were smoking (nonabstinent) at 12 months evidenced significantly higher RTs (M = 818.83 ms, SD = 120.91) relative to those who were abstinent (M = 692.54 ms, SD = 92.96), t(26) = 2.86, p = .008 (seeFigure 2).

Figure 2. Reaction time displayed by 12-month smoking cessation outcome (abstainers vs. nonabstainers). Error bars represent standard error of the mean.

Finally, to derive a better sense of the predictive utility of the RT measure (and based on the conceptual and statistical associations between RT and depressive symptoms), we conducted a stepwise binary logistic regression analysis, in which baseline depression (BDI–T1) served as the predictor variable in the first block, followed by RT entered in the next block; 12-month smoking cessation outcome was the dependent variable. Whereas analyses of the model with BDI alone revealed reasonable goodness-of-fit, Hosmer and Lemeshow test, χ²(6, N = 28) = 4.51, p = .61, the omnibus test of model coefficients was only marginally significant, χ²(1, N = 28) = 2.90, p = .09 (Nagelkerke R² = .14). Further, the odds ratio yielded by the BDI was 0.86, with an overall classification rate of 64% of participants correctly classified as either abstinent (4/10, 40%) or nonabstinent (14/18, 78%). Inclusion of RT in the model also yielded an adequate goodness-of-fit index, Hosmer and Lemeshow test, χ²(7, N = 28) = 10.90, p = .14. However, with the addition of RT in the model, the omnibus test reached significance, χ²(2, N = 28) = 8.51, p < .02 (Nagelkerke R² = .36). The overall classification rate reached 75%, with 83% of nonabstinent (15/18) and 60% of abstinent (6/10) individuals correctly classified (odds ratios of .96 and .97 for the BDI and RT measures, respectively). Moreover, univariate tests showed that whereas RT was a significant predictor of smoking cessation outcome (Wald = 3.89, p < .05), the BDI was not (Wald = 0.12, p > .70).

It is well established that there are strong and reliable associations between depression and smoking (Kassel & Hankin, 2006). Moreover, this relationship appears particularly strong in the context of smoking relapse; those smokers who are currently depressed, or even symptom-free but present with a history of depression, may be at greater risk for relapse subsequent to cessation (Wilhelm, Wedgwood, Niven, & Kay-Lambkin, 2006). Acknowledging that these data are based on measures of depressive symptomatology almost entirely derived from clinical interview and self-report, we hoped to improve on this situation by incorporating a measure of simple reaction time in the current study. The rationale here was that neurocognitive slowing has emerged as a potent marker of depressive symptomatology in and of itself (e.g., Pardo et al., 2006; White et al., 1997) and hence could potentially add discriminant validity to the prediction of smoking cessation outcomes.

As anticipated, we found that RT was positively correlated with depressive symptomatology, as measured by the BDI. Thus, both measures appear to be tapping shared aspects of depression. More important, however, RT was found to independently predict 12-month smoking cessation outcomes. Furthermore, this finding was derived from a sample of smokers who were not currently depressed but rather shared a history of at least one major depressive episode. Also of note was the observation that, whereas RT did predict cessation outcome, the BDI did not. Partial correlational analysis revealed that even when statistically controlling for the potential influence of the BDI, the association between RT and smoking status 12 months later held up. Hence, RT appears to be tapping an aspect of depression critical to its influence on smoking cessation. As such, future research needs to explore the mechanisms underlying these links.

Several limitations of the present need to be acknowledged. First, the sample size was relatively small. Thus, the extent to which our findings can be replicated with larger samples needs to be determined. Second, the sample was restricted to smokers with a history of major depressive disorder, none of whom met criteria for current major depression. Therefore, it remains to be seen whether slowed RT serves as a vulnerability factor for relapse among smokers without a history of depression or for smokers who enter treatment while in the throes of a major depressive episode. Correspondingly, it is conceivable that RT may actually have served as a proxy for some broader or related constructs, distinct from depression, for example, distress tolerance (Brown, Lejuez, Kahler, Strong, & Zvolensky, 2005) or task persistence (Brandon et al., 2003). Hence, in the absence of a control group comprising smokers without a history of depression, it becomes difficult to definitively attribute the observed RT differences to neurovegetative aspects of depression. Third, we used only one measure to assess neurocognitive functioning. Whereas it has been argued that depressive symptoms are associated with task-independent cognitive slowing (Pardo et al., 2006), future research should nonetheless further explore the association observed in this study by utilizing different types of cognitive tasks to see if, indeed, more specific cognitive deficits are predictive of cessation outcomes.

Finally, it is important to note that whereas the finding that RT predicts smoking status 12 months later is clearly important and, as far as we are aware, the first of its kind, more work needs to be done with respect to better understanding the underlying mechanisms that govern depression–smoking associations (see Kassel and Hankin, 2006, and Kassel et al., 2003, for a discussion of these issues). The present study demonstrated that a distal predictor influenced outcome many months later, but precisely how this relationship operates at the level of the individual smoker is as yet unknown. Indeed, delineation of within-subject, dynamic processes that contribute to smoking relapse (and relapse to other drugs and addictive behaviors as well) awaits future scrutiny (Shiffman, 2005).

In sum, this study suggests that one particular manifestation of depression—neurocognitive retardation as measured by reaction time on a simple decision task—accurately predicted treatment outcome 1 year later for 75% of our sample. Though these results need to be replicated and extended in future studies, one implication of our findings is that an RT measure could serve as a quick and cost-efficient method by which to assess aspects of depressive symptomatology not tapped by self-report measures. Indeed, the importance of fully understanding the link between smoking relapse and depression cannot be overstated. In recent years, virtually all state-of-the-art smoking cessation programs have come to incorporate some aspect of cognitive–behavioral therapy specifically aimed at reducing depressive symptoms (Wilhelm et al., 2006). Moreover, pharmacotherapy in the form of antidepressant medication has emerged as a first-line, effective approach to treating the smoker who desires to quit (Fiore et al., 2000). Because today's population of smokers is hardening, according to some researchers, and presenting with comorbid psychopathology, such as depression, it is critical that tobacco researchers continue to clarify the relationship between smoking and depression, as future lives are clearly at stake.

¹ Thirty consecutive study participants were asked to partake in this supplemental study, of whom 2 declined.

² Whereas the overall treatment outcome study from which this sample was derived comprised two treatment conditions (see Brown et al., 2001), no main effects for treatment were found in the full sample. Nonetheless, we assessed whether treatment condition influenced any of the findings reported in the present study; in all instances, inclusion of treatment condition did not change the reported findings.

³ During the RT task, errors to the neutral words were exceedingly rare, occurring, on average, less than once per participant. As such, there were no response speed–accuracy trade-off effects for RTs to the neutral words (r = −.05, p > .80).

⁴ The overall classification rate of 75% should be understood in context such that, even in the absence of any independent variables, the correct classification rate approaches 64%. Hence, RT ultimately allows for the correct prediction of 3 more individuals than would be seen using no predictors at all.

Anda, R. F., Williamson, D. F., Escobedo, L. G., Mast, E. E., Giovino, G. A., & Remington, P. L. (1990). Depression and the dynamics of smoking. Journal of the American Medical Association, 264, 1541–1545.

Beck, A. T., Ward, C. H., Mendelson, M., Mock, J., & Erbaugh, J. (1961). An inventory for measuring depression. Archives of General Psychiatry, 4, 561–571.

Brandon, T. H., Herzog, T. A., Juliano, L. M., Irvin, J. E., Lazev, A. B., & Simmons, V. N. (2003). Pretreatment task persistence predicts smoking cessation outcome. Journal of Abnormal Psychology, 112, 448–456.

Breslau, N., Kilbey, M. M., & Andreski, P. (1991). Nicotine dependence, major depression, and anxiety in young adults. Archives of General Psychiatry, 48, 1069–1074.

Brown, R. A., Kahler, C. W., Niaura, R., Abrams, D. B., Sales, S. D., Ramsey, S. E., et al. (2001). Cognitive–behavioral treatment for depression in smoking cessation. Journal of Consulting and Clinical Psychology, 69, 471–480.

Brown, R. A., Lejuez, C. W., Kahler, C. W., Strong, D. R., & Zvolensky, M. J. (2005). Distress tolerance and early smoking lapse. Clinical Psychology Review, 25, 713–733.

Cohen, S., Lichtenstein, E., Prochaska, J. O., Rossi, J. S., Gritz, E. R., Carr, C. R., et al. (1989). Debunking myths about self-quitting: Evidence from 10 prospective studies of persons who attempt to quit smoking by themselves. American Psychologist, 44, 1355–1365.

Covey, L. S. (1999). Tobacco cessation among patients with depression. Primary Care, 26, 691–706.

Covey, L. S., Glassman, A. H., & Stetner, F. (1997). Major depression following smoking cessation. American Journal of Psychiatry, 154, 263–265.

Covey, L. S., Glassman, A. H., & Stetner, F. (1998). Cigarette smoking and major depression. Journal of Addictive Diseases, 17, 35–46.

Degenhardt, L., & Hall, W. (2001). The relationship between tobacco use, substance-use disorders and mental health: Results from the National Survey of Mental Health and Well-being. Nicotine and Tobacco Research, 3, 225–234.

Fiore, M. C., Bailey, W. C., Cohen, S. J., Dorfman, S. F., Gritz, E. R., Heyman, R. B., et al. (2000). Treating tobacco use and dependence. Clinical practice guideline. Rockville, MD:U.S. Department of Health and Human Services, Public Health Service.

Glassman, A. H., Covey, L. S., Dalack, G. W., Stetner, F., Rivelli, S. K., Fleiss, J. F., et al. (1993). Smoking cessation, clonidine, and vulnerability to nicotine among dependent smokers. Clinical Pharmacology and Therapeutics, 54, 670–679.

Glassman, A. H., Helzer, J. E., Covey, L. S., Cottier, L. B., Stetner, F., Tipp, J. E., et al. (1990). Smoking, smoking cessation and major depression. Journal of the American Medical Association, 264, 1546–1549.

Haas, A. L., Munoz, R. F., Humfleet, G. L., Reus, V. I., & Hall, S. M. (2004). Influences of mood, depression history, and treatment modality on outcomes in smoking cessation. Journal of Consulting and Clinical Psychology, 72, 563–570.

Hall, S. M., Munoz, R. F., Reus, V. I., & Sees, K. L. (1993). Nicotine, negative affect, and depression. Journal of Consulting and Clinical Psychology, 61, 761–767.

Heatherton, T. F., Kozlowski, L. T., Frecker, R. C., & Fagerstrom, K. O. (1991). The Fagerstrom Test for Nicotine Dependence: A revision of the Fagerstrom Tolerance Questionnaire. British Journal of Addiction, 86, 1119–1127.

Hitsman, B., Borrelli, B., McChargue, D. E., Spring, B., & Niaura, R. (2003). History of depression and smoking cessation outcome: A meta-analysis. Journal of Consulting & Clinical Psychology, 71, 657–663.

Hughes, J. R., Gulliver, S. B., Fenwick, J. W., Valliere, W. A., Cruser, K., Pepper, S., et al. (1992). Smoking cessation among self-quitters. Health Psychology, 11, 331–334.

Hughes, J. R., Keely, J. P., Niaura, R. S., Ossip-Klein, D. J., Richmond, R. L., & Swan, G. E. (2003). Measures of abstinence in clinical trials: Issues and recommendations. Nicotine & Tobacco Research, 5, 13–25.

Jamner, L. D., Whalen, C. K., Loughlin, S. E., Mermelstein, R. J., Audrain-McGovern, J., Krishnan-Sarin, S., et al. (2003). Tobacco use across the formative years: A road map to developmental vulnerabilities. Nicotine & Tobacco Research, 5, S71–S87.

Kassel, J. D., & Hankin, B. L. (2006). Smoking and depression. In A.Steptoe (Ed.), Depression and physical illness (pp. 321–347). Cambridge, England: Cambridge University Press.

Kassel, J. D., Stroud, L. R., & Paronis, C. A. (2003). Smoking, stress, and negative affect: Correlation, causation, and context across stages of smoking. Psychological Bulletin, 129, 270–304.

Kassel, J. D., Weinstein, S., Skitch, S. A., Veilleux, J., & Mermelstein, R. (2005). The development of substance abuse in adolescence. In B. L.Hankin & J. R.Abela (Eds.), Development of psychopathology: A vulnerability-stress perspective (pp. 355–384). Thousand Oaks, CA: Sage.

Lasser, K., Boyd, J. W., Woolhandler, S., Himmelstein, D. U., McCormick, D., & Bor, D. H. (2000). Smoking and mental illness: A population-based prevalence study. Journal of the American Medical Association, 284, 2606–2610.

Niaura, R., Britt, D. M., Shadel, W. G., Goldstein, M., Abrams, D., & Brown, R. (2001). Symptoms of depression and survival experience among three samples of smokers trying to quit. Psychology of Addictive Behaviors, 15, 13–17.

Pardo, J. V., Pardo, P. J., Humes, S. W., & Posner, M. I. (2006). Neurocognitive dysfunction in antidepressant-free, non-elderly patients with unipolar depression: Alerting and covert orienting of visuospatial attention. Journal of Affective Disorders, 92, 71–78.

Piasecki, T. M. (2006). Relapse to smoking. Clinical Psychology Review, 26, 196–215.

Porter, R. J. (2003). Neurocognitive impairment in drug-free patients with major depressive disorder. British Journal of Psychiatry, 182, 214–220.

Ryan, C., Russo, K., & Greeley, J. (1996). Testing the global-slowing hypothesis: Are alcohol's effects on human performance process-specific or task-general?Acta Psychologica, 92, 59–78.

Shiffman, S. (2005). Dynamic influences on smoking relapse process. Journal of Personality, 73, 1715–1748.

Spitzer, R. L., Williams, J. B., Gibbon, M., & First, M. B. (1990). Structured Clinical Interview for DSM-III-R (nonpatient ed., Version 1.0). Washington, DC: American Psychiatric Press.

U.S. Department of Health and Human Services. (1990). The health benefits of smoking cessation. A report of the Surgeon General (DHHS Publication No. CDC 90–8416). Atlanta, GA: Public Health Service, Centers for Disease Control, National Center for Chronic Disease Prevention and Health Promotion, Office on Smoking and Health.

Warner, K. E., & Burns, D. M. (2003). Hardening and the hard-core smoker: Concepts, evidence, and implications. Nicotine & Tobacco Research, 5, 37–48.

White, D. A., Myerson, J., & Hale, S. (1997). How cognitive is psychomotor slowing in depression? Evidence from a meta-analysis. Aging, Neuropsychology, and Cognition, 4, 166–174.

Wilhelm, K., Wedgwood, L., Niven, H., & Kay-Lambkin, F. (2006). Smoking cessation and depression: Current knowledge and future directions. Drug and Alcohol Review, 25, 97–107.

Williams, J. M. G., Mathews, A., & MacLeod, C. (1996). The emotional Stroop task and psychopathology. Psychological Bulletin, 120, 3–24.

Submitted: May 31, 2006 Revised: November 7, 2006 Accepted: November 8, 2006

Between- and Within-Person Associations Between Negative Life Events and Alcohol Outcomes in Adolescents With ADHD

Acknowledgement: This research was principally supported by grants from the National Institute on Alcohol Abuse and Alcoholism: AA011873, AA007453, and AA00202. Additional support was provided by DA12414, MH50467, MH12010, ESO5015, AA12342, DA016631, MH065899, KAI-118-S1, DA85553, MH077676, MH069614, MH62946, MH53554, MH069434, IES LO3000665A, IESR324B060045, and NS39087. Sarah L. Pedersen was supported on grants from the National Institute on Alcohol Abuse and Alcoholism (K01AA021135) and ABMRF/The Foundation for Alcohol Research.

These hypotheses and analyses have not been presented in any form prior to this article.

Alcohol use disorder is most prevalent between ages 18 and 29 years (Grant et al., 2015), but the development (e.g., the initiation of use and the first appearance of problems) of these disorders begins earlier in adolescence (Meich, Johnston, O’Malley, Bachman, & Schulenberg, 2015). As the average quantity and frequency of alcohol use increases in the general population from adolescence into young adulthood, variability in alcohol use also steadily increases with age, with some adolescents escalating rapidly into heavy drinking and eventually problems (e.g., Hussong, Bauer, & Chassin, 2008), whereas others remain light or moderate drinkers. Only a fraction of those who begin drinking in adolescence eventually develop an alcohol use disorder, with prevalence estimates of the past year of any alcohol use disorder between ages 18 and 29 years at approximately 26% (Grant et al., 2015). As such, understanding what influences the developmental trajectories of alcohol involvement across adolescent development remains a priority of research.

Although a large literature has outlined the pathways by which externalizing behaviors may influence the development of adolescent alcohol involvement (see review by Chassin, Colder, Hussong, & Sher, 2016), emerging evidence suggests that exposure to negative life events may independently shape alcohol use trajectories and presage worsening outcomes (Keyes, Hatzenbuehler, Grant, & Hasin, 2012). Negative life events predict higher levels of alcohol use, alcohol-related problems, and alcohol use disorders among adolescents and young adults (Cerbone & Larison, 2000; King & Chassin, 2008; Wills, Sandy, & Yaeger, 2002). Prospective studies that have focused on change over time in alcohol use suggested that exposure to negative life events was associated with escalating trajectories of alcohol use during adolescence (King, Molina, & Chassin, 2009; Wills, Sandy, Yaeger, Cleary, & Shinar, 2001).

This prior research on the negative life event–alcohol use association focused on between-person associations, in which those who report more negative life events at an earlier time point exhibited, on average, higher levels or greater increases in alcohol use and problems over time. However, most hypotheses about the role of negative life events in alcohol use focus on within-person processes, hypothesizing that alcohol use occurs when an individual is both exposed to negative life events and utilizes maladaptive coping strategies in the face of those negative life events (Chassin et al., 2016; Sher, Grekin, & Williams, 2005). Some individuals are more likely to experience negative life events, either because of contextual or individual factors such as parenting, temperament, or socioeconomic status (King, Molina, & Chassin, 2008). Thus, it is important to disaggregate the between-individual effects of exposure, which may reflect more stable individual differences in the propensity to experience negative life events, from the within-individual effects of the life events themselves, which may better represent the process of stress adaptation as well as other time-varying factors that influence both stress and drinking (Curran & Bauer, 2011).

To date, one study showed that time-varying differences in negative life events were related to time specific increases (i.e., those not accounted for by an adolescent’s average trajectory of use) in alcohol use and binge drinking during adolescence (Aseltine & Gore, 2000), but that study did not explicitly separate the within- and between-person associations of life events with alcohol use (Enders & Tofighi, 2007). Using a latent growth curve modeling framework, a second study showed that family life events were related to between- and within-person differences in alcohol use during adolescence (King et al., 2009). However, more recent methodological studies have suggested that some of the methods of that study, such as binning ordinal measures of alcohol use frequency (McGinley & Curran, 2014) or failing to disaggregate between- from within-person variance as predictors in growth models (Curran, Howard, Bainter, Lane, & McGinley, 2014), may have inflated the time-varying associations in those models. Finally, neither prior study accounted for the heavily skewed and zero-inflated nature of adolescent alcohol use (Atkins, Baldwin, Zheng, Gallop, & Neighbors, 2013).

Attention-Deficit Hyperactivity Disorder (ADHD), Negative Life Events, and Alcohol Use

Substantial evidence suggests that adolescents with ADHD are at a heightened risk of experiencing negative life events: ADHD is associated with academic and social difficulties that may directly increase the likelihood of experiencing negative life events (e.g., getting bad grades, fighting with peers [Barkley, 2006]). Moreover, many children with ADHD continue to meet diagnostic criteria in adolescence (e.g., Barkley, Murphy, & Fischer, 2008), and even more experience impairment despite subclinical levels of ADHD symptoms (Sibley et al., 2012). Continued ADHD-related impairments may also indirectly increase the experience of other negative life events (e.g., parental decisions to restrict activities or resources). Finally, children and adolescents with ADHD come from families in which they are exposed to higher levels of negative life events, such as exposure to marital conflict, divorce, general family adversity, and parental alcoholism (Counts, Nigg, Stawicki, Rappley, & von Eye, 2005; Knopik et al., 2006; Wymbs et al., 2008). It may be that adolescents with a diagnosis of ADHD report higher levels of negative life events because of the downstream effects of early ADHD (such as diminished peer relations), the background variables associated with a diagnosis of ADHD (such as family problems), or the effects of continued impairment from ADHD.

Adolescents with ADHD not only may report greater exposure to negative life events, but they may also be especially sensitive to their effects. Their diminished skills for coping with distress (Hampel, Manhal, Roos, & Desman, 2008; Molina, Marshal, Pelham, & Wirth, 2005), among other vulnerability characteristics such as weak executive function and generally increased impulsivity, support this hypothesis. One study from our group found a stronger cross-sectional association between academic life events (e.g., doing poorly on an exam) and problem alcohol use for adolescents with, versus without, ADHD histories (Marshal, Molina, Pelham, & Cheong, 2007). Relatedly, other widely studied environmental factors, such as peer alcohol use (Marshal, Molina, & Pelham, 2003) and parental monitoring (Walther et al., 2012), have been shown to be more strongly related to alcohol use for those with ADHD compared with those without ADHD. Understanding the potential for stress vulnerability in ADHD is particularly important, given the increased risk for alcohol use disorder that characterizes this population in adulthood (e.g., Lee, Humphreys, Flory, Liu, & Glass, 2011).

Whereas much prior research on negative life event exposure and alcohol use has utilized simple count measures of life events (which require an adolescent to report whether an event occurred), research on negative life events highlights the importance of considering individuals’ perception of events (Grant et al., 2003). For example, Davis and Compas (Davis & Compas, 1986) found that the desirability of a life event (whether rated as negative or positive) was positively related to students’ perception that they could cope with events (r = .84). Moreover, the mere number of stressors and how stressors are perceived may explain different variation in psychopathology (Duggal et al., 2000). Classic models of stress and psychopathology argue that stressors should be perceived as a challenge or threat to the individual (Lazarus & Folkman, 1984), although reviews of the stress literature show that both counts of stressors and an individuals’ perception of those stressors can be useful (Grant et al., 2004). To date, most research has relied on counts of stressors and has not considered whether an adolescent’s perception of stressors explains variance in alcohol use. Adolescents with ADHD are known to overestimate their competence and underestimate their impairment across a number of life domains (Evangelista, Owens, Golden, & Pelham, 2008; Hoza et al., 2004). It may be that adolescents with ADHD report a higher number of life events but (due to cognitive biases) report that they are less impactful or upsetting and that their perception of events differentially alters the relation between negative life events and alcohol outcomes.

Current Study

The current study extends prior research by examining the between- and within-person associations of negative life events and alcohol outcomes in a sample of adolescents with and without a well-established diagnosis of ADHD during childhood, using statistical methods that better account for the skewed and zero-inflated nature of those outcomes. The main goal of the current study was to replicate prior work while attempting to explicitly address the methodological challenges raised in recent studies as well as extending these models to a new high-risk sample, adolescents with ADHD. We hypothesized that both between- and within-individual differences in exposure to negative life events would be associated with alcohol use. The second goal of the current study was to test whether children diagnosed with ADHD reported greater numbers of negative life events during adolescence and to examine the contribution of concurrent ADHD symptoms to negative life events. Third, we aimed to test whether adolescents with ADHD showed stronger associations between life events and alcohol behaviors during adolescence. We hypothesized that ADHD history would predict the experience of negative life events and, most importantly, strengthen the association between life events and alcohol outcomes. Finally, we compared simple counts of negative life events with the adolescent’s perception of the negativity of those events.

Participants

More detailed information on the recruitment of the Pittsburgh ADHD Longitudinal Study (PALS) may be found in another report (Molina et al., 2012).

ADHD group

Participants with childhood ADHD were diagnosed with Diagnostic and Statistical Manual of Mental Disorders III, revised (DSM–III–R) or DSM–IV ADHD in childhood, at an average age of 9.40 years (SD = 2.27). Participants with ADHD were selected for longitudinal follow-up with annual interviews because of their diagnosis of ADHD and participation in a summer treatment program for children with ADHD, an 8-week intervention that included behavioral modification, parent training, and psychoactive medication trials where indicated (Pelham & Hoza, 1996).

Participants with ADHD were assessed in childhood using standardized parent and teacher DSM–III–R and DSM–IV disruptive behavior disorder symptom rating scales (Pelham, Gnagy, Greenslade, & Milich, 1992) and a standardized semistructured diagnostic interview administered to parents by a PhD-level clinician. Two PhD-level clinicians independently reviewed all ratings and interviews to confirm DSM diagnoses, and when disagreement occurred, a third clinician reviewed the file and the majority decision was used. Exclusion criteria for follow-up was assessed in childhood and included a full-scale intelligence quotient <80, a history of seizures or other neurological problems, and/or a history of pervasive developmental disorder, schizophrenia, or other psychotic or organic mental disorders. At the first PALS follow-up interview, which occurred on a rolling basis between 1999 and 2003, the mean age was 17.75 years (SD = 3.39 years, range = 11–25).

Non-ADHD group

Adolescents without ADHD were recruited into the PALS when those with ADHD were recruited for follow-up. Non-ADHD comparison participants were recruited on a rolling basis to ensure demographic similarity to the ADHD group (age within 1 year, sex, race, highest parental education) and were recruited from the same regional area as the participants with ADHD. Individuals who met DSM–III–R criteria for ADHD (presence of eight or more symptoms reported by either the parent or young adult participant), currently or historically, were excluded. Non-ADHD comparison participants with subthreshold ADHD symptomatology, or with other psychiatric disorders, were retained.

Procedure

Interviews for the PALS were conducted annually in adolescence. Interviews were conducted in the ADD Program offices by postbaccalaureate research staff. Informed consent was obtained and all participants were assured confidentiality of all disclosed material except in cases of impending danger or harm to self or others (reinforced with a Department of Health and Human Services Certificate of Confidentiality). In cases in which distance prevented participant travel to the research offices, information was collected through a combination of mailed and telephone correspondence; home visits were offered as need dictated. Self-report questionnaires were completed either with paper and pencil or Web-based versions on a closed circuit Internet page. All procedures were approved by the Institutional Review Board of Western Psychiatric Institute and Clinic.

Selection of the Current Sample

Data were selected from the first four annual interviews of adolescents based on procedures used elsewhere to test longitudinal hypotheses about adolescent functioning (Molina et al., 2012). Participants were selected if they were interviewed one or more times between the ages of 14 and 17 years. Because multilevel modeling and generalized estimating equations make use of all available data at Level 1 (Raudenbush & Bryk, 2002), participants were excluded only if they were not interviewed between ages 14 and 17 years. For the resulting subsample (n = 259), there were no statistically significant differences between the ADHD (n = 146) and non-ADHD (n = 113) groups on sex or ethnic/racial minority but a statistically significant difference for highest parental education and household income (lower in the ADHD group). For analysis, data were organized by age at interview to allow modeling of life events and alcohol use longitudinally by age (Bollen & Curran, 2006). This provided data for life events and alcohol use at one (n = 43), two (n = 79), three (n = 86), or four (n = 51) occasions. Participants provided data at ages 14 (n = 114), 15 (n = 158), 16 (n = 166), and 17 years (n = 167). To estimate between-person associations, we had data from 259 participants with 756 observations. To estimate within-person associations, we had data from 216 participants (with 689 repeated observations), 129 of whom (with 388 observations) reported any alcohol use. Table 1 provides descriptive statistics for the current sample.

Demographics of Current Sample (n = 259)

Measures

Background variables: Parental characteristics, sex, and race

Because they have been shown to influence the occurrence of life events, alcohol use, or both, for all analyses we initially controlled for the baseline presence of a parental alcohol use disorder, parental antisociality, maternal depressive symptoms, parental divorce, sex, and race. Parental alcohol use disorder was coded as present if either parent met criteria on the Structured Clinical Interview for DSM for Nonpatients (Spitzer, Williams, Gibbon, & First, 1990) by their own report or was reported by the other parent on the Michigan Alcoholism Screening Test–Short (Selzer, Vinokur, & van Rooijen, 1975). The parent on the Michigan Alcoholism Screening Test–Short focuses on consequences from problematic drinking, and a score of 3 or higher was coded as having an alcohol use disorder. These two assessments were combined and coded as 1, either parent met criteria based on self/other-report, vs. 0, neither parent met criteria based on self/other-report. Parental antisociality was coded as present if either parent met criteria on the Structured Clinical Interview for DSM for Nonpatients (Spitzer et al., 1990). Maternal depressive symptoms were assessed by maternal self-report on the 21-item Beck Depression Inventory (Beck, Steer, & Carbin, 1988). Sex was self-reported (0, female, 1, male), as was race (0, White, 1, non-White).

Negative life events

Negative life events in the past year were assessed annually with 120 items from the Adolescent Perceived Events Scale (Compas, Davis, Forsythe, & Wagner, 1987). The Adolescent Perceived Events Scale has been used extensively in prior research and has been shown to predict both internalizing and externalizing symptoms in adolescents (Grant et al., 2003; McMahon, Grant, Compas, Thurm, & Ey, 2003). For each item, adolescents indicated whether an event occurred and, using a 9-point scale, the degree to which it was experienced as negative or positive (1, extremely bad, to 9, extremely good). Example life events were parents getting divorced, parent loses a job, having few or no friends, not getting along with parents of friends, doing poorly on an exam or paper, problems or arguments with teachers or principal, getting in trouble or being suspended from school, death of a family member, change in the health of a friend, and hospitalization of a family member or relative. We excluded 31 items from the original scale that assessed minor life events (such as going to church/synagogue or helping other people), psychological symptoms, bereavement/illness (which also occurred very rarely), substance use, or ADHD diagnosis or treatment, leaving 89 total negative life events. Following prior work with this scale (Wagner & Compas, 1990), we computed both a count of negative life events and a score reflecting the subjective evaluation of those negative events. Prior research has indicated that both the accumulation of life events as well as the adolescent’s perception of them are independent predictors of psychopathology (Grant, Compas, Thurm, McMahon, & Gipson, 2004). The count was comprised of all items rated by the adolescent as at least slightly bad (4) to extremely bad (1). The subjective evaluation score was computed as the mean of the adolescent’s ratings of all events that were rated as at least slightly bad after reverse scoring the ratings (e.g., extremely bad [4]). Across age, these two scores were correlated very weakly, r = .18, p < .001).

Table 1 provides descriptive statistics for negative life events. On average, adolescents reported approximately 11 different negative life events in the past year (range = 0–58) and reported an average perception of 2.36, which represents a response of somewhat bad.

ADHD symptoms

Childhood diagnosis of ADHD is described above. At each wave, ADHD symptoms were measured using parent and teacher report of 18 DSM–IV ADHD symptoms (Pelham, Gnagy, Greenslade, & Milich, 1992), scored on a scale of 0 (not at all) to 3 (very much). We then took the maximum score across the two raters for each symptom and computed a mean across all symptoms at each age.

Alcohol Use

Alcohol use was assessed at each annual interview with a structured paper-and-pencil substance use questionnaire (Molina & Pelham, 2003; Molina, Pelham, Gnagy, Thompson, & Marshal, 2007). The substance use questionnaire is an adaptation of existing measures, including the Health Behavior Questionnaire (Jessor, Donovan, & Costa, 1989) and the National Household Survey on Drug Abuse interview (Substance Abuse and Mental Health Services Administration, 1992) and includes both lifetime exposure questions (e.g., have you ever had a drink, age of first drink) and quantity/frequency questions for alcohol and other substances. The current study utilized two items that assessed frequency of use and drunkenness over the past 12 months. Items used a 12-point scale (from never to several times a day). We tested alcohol outcomes separately because there are concerns in the literature about combining across alcohol outcomes and/or converting them to pseudocount variables (as we did previously; King et al., 2009), particularly when they are measured with an ordinal scale (McGinley & Curran, 2014).

Analytic Strategy

We were interested in predicting between-person differences in negative life events during adolescence from childhood and adolescent ADHD symptoms and in predicting within- and between-person differences in alcohol outcomes during adolescence from life events and ADHD. However, our alcohol outcomes were heavily skewed and zero-inflated, in that many adolescents did not report drinking, and when they did, most reported fairly low levels. This produces nonnormality in the residuals, violating the assumptions of multilevel models (MLMs) and can produce bad parameter estimates and misleading inferences (Atkins, Baldwin, Zheng, Gallop, & Neighbors, 2013). Thus, to predict alcohol outcomes, we used generalized estimation equations (GEEs; Zeger, Liang, & Albert, 1988), GEE readily allows the estimation of zero-inflated and hurdle count models (such as zero-inflated Poisson or hurdle negative binomial models), which may better estimate the response generation process for these variables while also accounting for the effects of clustering within the individual. MLMs and GEEs were well suited as analytic approaches because both estimate parameters using the available Level 1 data (i.e., repeated observations of individuals across age), do not require all Level 2 observations (i.e., participants) to have identical or balanced observations at Level 1 and readily allow the separation of between- and within-components of variance in predictors and outcomes (Raudenbush & Bryk, 2002).

We tested our hypotheses in R 3.2.3 using the nlme and pscl packages with the maximum likelihood estimator. We used MLMs to predict negative life events across adolescence and GEEs to predict alcohol outcomes. To estimate GEEs, we estimated generalized linear models with the appropriate link function (such as negative binomial or hurdle negative binomial) using the r package pscl (Zeileis, Kleiber, & Jackman, 2008) and used a custom sandwich estimator to correct model standard errors for clustering (D. Hu, personal communication). This corrects the standard errors of the fixed effects for the effects of clustering while avoiding the problems that can arise from trying to model the random covariance structure (i.e., slope and intercept variability) of complex distributions such as zero-inflated count distributions. Because our hypotheses were related to fixed effects only (such as estimating the associations between ADHD and the slope of alcohol use), rather than on obtaining a population level estimate of individual differences in slopes, GEEs were an appropriate choice to test the current hypotheses.

Zero-inflated hurdle models

Although alcohol outcomes were measured on an ordinal scale, we utilized models for count data because they best fit the distributions of the alcohol variables. Across all alcohol outcomes, model fit indices (Bayesian information criterion and Akaike information criteria) suggested that zero-inflated or hurdle-negative binomial models best fit the data; we chose hurdle negative binomial models because they best fit our interpretation of adolescents’ actual behavior. Hurdle negative binomial models separately model the presence or absence of the outcome (i.e., the hurdle, or likelihood) and, among those with any level of the outcome, the count (or level) of the outcome as a negative binomial distribution, which has a variance that is greater than its mean (Hilbe, 2011). Thus, for each outcome, coefficients predicting the likelihood of the outcome occurring (i.e., whether or not an adolescent reported drinking in the past year) may be transformed into an odds ratio (OR), which predicts the relative odds of an event occurring. Coefficients predicting the level of an outcome are converted to a rate ratio (RR), which predicts the number of events (such as the number of drinks in the past year that an adolescent may report).

Centering of life events

To disaggregate the between- and within-person associations of negative life events with alcohol outcomes, we used a combination of centering within cluster at Level 1 and grand-mean centering at Level 2 (Enders & Tofighi, 2007), which perfectly separates variation in a given predictor into within- and between-person variability. Centering within cluster is achieved by subtracting a participant-level mean across observations from each participant’s score at each time point. This provides a time-specific score that reflects only within-person variance, and observations at each time point essentially become a deviation score, representing that person’s deviation from their own average at that time point. The participant’s mean score across observations may be grand-mean centered by subtracting each participant’s mean from the sample average of all participant means, which can then be entered as their between-person variable. This score represents a participant’s average deviation from the sample mean and reflects their average level of life event exposure across adolescence. These resulting centering within cluster and grand-mean centered scores are perfectly uncorrelated (r = .00) because they partial within- and between-person variance in life event exposure over time. In this way, a multilevel model may be utilized to address state-trait questions by a simple centering scheme.

Model-fitting approach

We followed a standardized approach to model fitting. For all model comparisons, we relied on Akaike information criteria and Bayesian information criterion as tests of relative model fit (Raftery, 1995) prior to applying the GEE correction to account for clustering within subjects. We first tested for the general shape of change for both alcohol outcomes, comparing linear and quadratic models. Because there was little variability in alcohol outcomes at age 14 years, we used age 17 years as the intercept and estimated growth from age 14 to 17 years, as in our previous research (Molina et al., 2012). To ensure that the main hypothesis tests were not biased by unmodeled dependencies in the data, we tested all covariate by predictor interactions as well as the between- by within-person effects of both count and perceptions of life events. This is recommended as best practice for model building in regression models (Allison, 1977), and simulations have shown that not including or estimating interactions that exist in models can induce substantial bias in the main effects coefficients (Vatcheva, Lee, McCormick, & Rahbar, 2015).

To balance the risk of alpha inflation against model misspecification, we used an a priori threshold of p < .01 to retain significant covariate by predictor interactions and refrained from interpreting any interactions we did retain to avoid speculation about nonhypothesized interactions. Then we examined the main effects of between- and within-person negative life events on each outcome. Next, we tested whether childhood ADHD was associated with life events during adolescence and compared those models with ones measuring the effects of concurrent ADHD symptoms. Finally, we tested whether childhood diagnosis and concurrent symptoms of ADHD moderated the between- and within-person associations of life events with each alcohol outcome. Again, because we had only general hypotheses about this effect (i.e., that life events would have a stronger relation for those with ADHD or more ADHD symptoms), we used the Benjamini-Hochberg correction (Benjamini & Hochberg, 1995) to control for the false discovery rate. We initially controlled for parental alcoholism, antisociality, and depression in all analyses, but controlling for them did not change the magnitude of the coefficients or inferences from the final models, so we dropped them for the sake of parsimony. Thus, the final models controlled for sex, race, and parental divorce to control for baseline differences.

Descriptive Statistics

Adolescents with and without childhood diagnoses of ADHD reported similar counts and perceptions of life events (all t(277) < 1.28, p > .20), averaging approximately 11 negative life events and reporting them to be somewhat bad on average. Among those with a childhood ADHD diagnosis, the average ADHD symptom score from age 14 to 17 years was 1.51 (SD = .63), whereas the average ADHD symptom score for those without a childhood diagnosis was lower (M = .48, SD = .37), t(277) = −15.42, p < .001.

Only two covariate-by-predictor interactions were significant at our a priori threshold of p < .01. A significant interaction of race (White vs. non-White) by divorce (nondivorced vs. divorced parents) (p = .0022) suggested that the effects of parental divorce on the likelihood of any alcohol use were smaller for non-White adolescents (OR = .05) than for White adolescents (OR = .60). Because these interactions were not hypothesized, we do not interpret them further, but we did include them in all further models to reduce model misfit and to improve coefficient estimation.

Unconditional Models of Alcohol Involvement

We illustrate the unconditional growth models in Figure 1 to aid interpretation. For all alcohol outcomes, a linear effect of time best fit the data, with all fit indices for the quadratic model greater than those for the linear models, suggesting the linear model of time fit better than the quadratic. Unconditional model results, with estimates of intercepts and slopes, are presented in Table 2. Generally, the likelihood of alcohol involvement across each outcome increased over time, with the odds of reporting any alcohol use or drunkenness increasing by 1.7–1.9 per year from age 14 to 17 years. In other words, for every year that passed, the likelihood of any alcohol use or drunkenness nearly doubled. For example, the probability of reporting any alcohol use rose from less than 20% at age 14 years to more than 50% by age 17 years, whereas the probability of reporting any drunkenness rose from less than 10% to around 35%.

Figure 1. Unconditional growth models of past year alcohol involvement. a, frequency of use; b, frequency of drunkenness. Model predicted probability and level of alcohol outcomes by age. Confidence intervals were simulated using the simcf package (C. Adolph, personal communication).

Unconditional Growth Model

There were also linear increases in the frequency of alcohol use over time, with drinking increasing by 18–19% per year (controlling for the influence of the covariates) among those who reported any alcohol use. On the other hand, those who reported any drunkenness exhibited no increases in the level over time, even as other adolescents became more likely to drink in that manner.

Negative Life Events and Alcohol Involvement

We then examined the between- and within-person associations of negative life events with alcohol use during adolescence, predicting the likelihood and level of alcohol outcomes from between- and within-person variability in the number and perception of life events.

The most consistent effect we observed was that between-person differences in the average count of negative life events over time were associated with higher age 17 year likelihoods of alcohol use (OR = 1.10) and drunkenness (OR = 1.069). Figure 2 illustrates the effects of life events on trajectories of the likelihood of alcohol use. Additionally, within-person variance in the number of negative life events was related to an increased likelihood of drunkenness (OR = 1.045) and a higher level (RR = 1.035) of alcohol use, meaning that in any given year, reporting more negative life events than expected was related to higher likelihoods of reporting drunkenness and more frequent alcohol use than what would be predicted by that adolescent’s own trajectory of use. Moreover, after correction for the false discovery rate, we observed an association of within-person variance in the number of negative life events and the likelihood of any alcohol use (OR = 1.033, p = .06).

Figure 2. Between-person differences in life event exposure predicts heightened probability of alcohol use over time. Model predicted probability of alcohol use by age at −1, mean and +1 SD of between-person count life event exposure. Confidence intervals were simulated using the simcf package (C. Adolph, personal communication).

On the other hand, the adolescent’s perception of those life events was not related to either drinking outcome at either the between- or within-person level.

Predicting Trait-Negative Life Events From ADHD

We next tested whether childhood ADHD predicted a larger number or worse perception of negative life events during adolescence using MLM. There was substantial between (54%) and within-person (46%) variability in adolescents’ report of both the number and perception of negative life events. None of the covariates were related to the perception or number of negative life events across adolescence. Moreover, only the effect of ADHD on the perception of negative life events approached significance (b = .152, SE = 0.082, p = .063), which suggested that adolescents with ADHD reported marginally more negative perceptions of negative life events from age 14 to 17 years but did not report any more or fewer negative life events relative to adolescents without ADHD.

This effect seemed to be confirmed in separate models using adolescent ADHD symptoms as a predictor of life events: The average level of ADHD symptoms across adolescence was associated with reporting more negative perceptions (but not higher numbers) of negative life events (b = .12, SE = .05, p = .035). In other words, higher levels of average ADHD symptoms across adolescence were related to more negative perceptions of life events on average.

Does ADHD Moderate the Effects of Life Events on Alcohol Involvement Across Adolescence?

Finally, we tested whether the relation between counts or perceptions of negative life events on alcohol use differed, depending on childhood ADHD diagnosis or adolescent ADHD symptoms. Table 3 presents these final results. We tested this hypothesis by including ADHD diagnosis or adolescent ADHD symptoms (in separate models) as predictors of level and change in alcohol involvement and as moderator(s) of the between- and within-person effects of life events described above. There were no main effects of childhood ADHD or moderation of life events by childhood ADHD that survived correction for the false discovery rate. This suggested that there was little support for the notion that childhood ADHD moderated the effects of life events on alcohol use.

Effects of Negative Life Events and ADHD on Alcohol Involvement

Between-person differences in ADHD symptoms during adolescence were associated with higher levels of alcohol use frequency (RR = 1.31) and drunkenness (RR = 1.53) at age 17 years. Among those who reported average or high levels of average ADHD symptoms across adolescence, the level of alcohol outcomes rose accordingly. There were no other main effects of ADHD symptoms during adolescence, nor did ADHD symptoms moderate the effects of life events on alcohol outcomes. Moreover, the associations between life events and alcohol use were largely unchanged with the inclusion of ADHD symptoms.

The goals of the current study were to extend prior research on the between- and within-person associations between negative life events and alcohol involvement during adolescence in a high-risk sample using a broad measure of life events and statistical models that better accounted for between- and within-person variability in life events as well as the nonnormal distributions of alcohol outcomes in adolescence. We tested whether ADHD was associated with heightened vulnerability to negative life event exposure and whether ADHD predicted a stronger association between negative life event exposure and alcohol involvement. Overall, our results largely suggested that the number, but not the perception, of negative life events was associated with both between- and within-person changes in the level or likelihood of alcohol involvement during adolescence. Only ADHD symptoms that persisted across adolescence were associated with more negative perceptions of life events (but not the number of life events) as well as with higher levels of alcohol use and drunkenness. Neither childhood diagnosis of ADHD nor persistence of ADHD symptoms strengthened the association between life event exposure and alcohol involvement.

Previous work (King et al., 2009) suggested that both between- and within-person exposure to uncontrollable stressors (familial life events) were related to trajectories of alcohol use. We partially replicated and extended this finding, showing that between-person differences in the number of negative life events were associated with an increased likelihood of alcohol involvement. Specifically, adolescents who reported an average number of negative life events that was 1 SD above the mean across adolescence also reported a likelihood of any drinking that was nearly twice as high (OR = 1.96; obtained by multiplying the model coefficient by the SD of between-person count of life events and then exponentiating) and a likelihood of getting drunk that was greater than 1.5 times as high (OR = 1.61) as an adolescent at the mean number of life events. Conversely, adolescents whose perceptions of life events were 1 SD more negative than average were no more likely to drink (OR = 1.04) or report getting drunk (OR = 1.05) than those whose perceptions were at the sample average. One interpretation of this finding is a third-variable explanation: Adolescents who are prone to experience negative life events, such as those with high levels of personality risk, or those in environmental contexts that expose them to high levels of adversity over time, are also more likely to drink and get drunk. Alternately, these findings may suggest that negative life events may be impactful because of their occurrence, rather than by their perception by the adolescent (Duggal et al., 2000). Altering appraisals of negative life events may be less effective than interventions that might seek to either reduce the number of negative life events themselves (by reducing controllable negative life events, such as by improving social skills) or by providing environmental supports that may counter the effects of uncontrollable negative life events whether or not an adolescent perceives them to be negative (e.g., increasing involvement in prosocial activities). Interestingly, most research and theory on interventions to address stress among youth emphasize improved individual’s coping or emotion regulation skills (Izard, 2002), but our findings lead us to speculate that supplemental approaches that counter the loss of resources that accompany accumulated negative life events (e.g., transportation to extramural activities needed after parental job loss; tutoring to raise poor grades) might be especially helpful. A growing literature on interventions that directly addresses ADHD-related impairments in adolescence is also relevant (Sibley et al., 2016). Future studies contrasting these approaches, and their associations with alcohol and other health risk behaviors, are warranted.

We also observed a relatively consistent within-person association: above and beyond the variance explained by age, when adolescents reported more negative life events than what was typical for them, in that same year they had higher likelihoods of drunkenness and higher levels of alcohol use that were not explained by their developmental trajectories of alcohol involvement (the association with the likelihood of alcohol use approached significance). Adolescents who reported life events in a year that were 1 SD above the average number of life events they reported across the study also reported 15% higher levels (RR = 1.15) of alcohol use, and 1.20 times the odds (OR = 1.20) of reporting drunkenness in that year, relative to the expected level and likelihood at their average number of life events. On the other hand, the relative associations of within-person fluctuations in perceptions with the level of alcohol use and the likelihood of drunkenness were much smaller and not significant (RR = .91, OR = .89, respectively). This finding extends our previous work (King et al., 2009) by showing that the associations of a broader range of negative life events beyond the relatively narrow range of family related life events captured by that earlier study are related to increased risk. Future interventions/preventions may directly benefit from these results. For example, alcohol prevention/intervention efforts could focus on adolescents who report a recent life event (e.g., such as parental divorce or school transitions) because this may be a time when alcohol use will subsequently increase. Moreover, our findings suggest that it is the events themselves, not the adolescents’ perception of them, that explain the within-person associations of negative life events with alcohol outcomes. Future research should explore the degree to which these state associations generalize to other forms of externalizing and internalizing psychopathology and whether other time varying factors (such as social support or coping skills) may moderate the associations of life events with psychopathology to guide the target of intervention. However, we should also caution that these within-person associations, which represent retrospective associations at the yearly level, cannot determine the true direction of effect; studies with a more time-sensitive design can bring us closer to an understanding of the connections between stress and alcohol use in the moment.

In general, adolescents with a childhood diagnosis of ADHD were no different in terms of their experience of the number or perception of negative life events between ages 14 and 17 years. However, between-person differences in current ADHD symptoms (i.e., parent- and teacher-reported symptoms during adolescence) were associated with a more negative perception of negative life events. Although a proliferation of studies have shown that adolescents and young adults with ADHD histories perceive less symptomatology and impairment than reported by their peers or parents (e.g., Mrug, Hoza, & Bukowski, 2004), this positive self-perception bias may not fully extend to perception of negative life events as stressful. This finding does not rule out the possibility that adolescents with ADHD actually underreport the occurrence of negative life events but perceive those that do occur as more negative. This differential finding may suggest that it is inattention to event occurrence that explains positive biases. Both symptoms of and impairment from ADHD for many adolescents continue to persist into adolescence (Barkley et al., 2008; Sibley et al., 2012), and these impairments may contribute to the perception of negative life events as more negative or stressful, particularly in the familial, school, and social domains as adolescents navigate the challenges of developing autonomy and individuation from parents, increasingly challenging school and social demands (Bagwell, Molina, Pelham, & Hoza, 2001; Kent et al., 2011). A number of studies have shown that symptom persistence in adolescence is associated with other externalizing problems such as oppositional defiant disorder and conduct disorder (i.e., Costello & Maughan, 2015), emotion problems, suicidality, and academic failure and dropout (Costello & Maughan, 2015; Kessler et al., 2014) as well as early adult substance use (Howard et al., 2015) including, in this sample, an association between ADHD symptom persistence, delinquency, and frequency of alcohol use (Molina et al., 2012, and replicated in the current study).

On the other hand, neither these life event perceptions, nor the experience of the negative life events themselves, were more strongly associated with alcohol outcomes for adolescents with a history of ADHD or with ongoing ADHD symptoms. These results conflict with prior work suggesting that adolescents with ADHD may be more susceptible to environmental conditions in regard to alcohol use (e.g., peer alcohol use, parenting factors, Belendiuk, Pedersen, King, Pelham, & Molina, 2016; Marshal et al., 2007; Walther et al., 2012). The differences between prior studies and the current one may reflect the longitudinal nature of the current study, the emphasis on the between- and within-individual differences in negative life events, or the treatment of alcohol use as a zero-inflated count outcome. It may also be important to consider additional dimensions of life event perception beyond positivity and negativity, such as the impact a life event has on an adolescent, or how important an adolescent views the event itself. Further research utilizing multiple informants of life events (e.g., school records, parental report) would also be useful as a check on the role of positive self-perception bias on our findings. In addition, a direct comparison of the effects of parent-reported impairments from ADHD, with the most typical being academic, behavioral, and social (Barkley et al., 2008), with those of self-reported negative life events associated with the experience of these impairments would further specify sources of alcohol use vulnerability for adolescents with ADHD. Mechanistic studies of negative affect, versus impairment-driven, pathways could follow (Molina & Pelham, 2014).

It should also be considered that both our null and significant findings could have been influenced by low power to detect effects in the current study, particularly at the within-person level, and that lack of support for certain effects (especially interactions) may at best suggest that effects that do exist may be smaller than the significant effects we were able to detect. Those effects were only estimable using data from individuals with more than one time point (n = 216 participants with 689 repeated observations, and only 129 of those participants reported any alcohol use). It is not well understood what study or design factors influence statistical power in GEEs, especially for hurdle count GEEs, and there are no guidelines for standard measures of effect size for count models in terms of what constitutes a small, medium, or large effect size. The relatively low variability in alcohol outcomes in the current study may have influenced our ability to detect associations, and thus, the null effects we report may not be reliably ruled out unless they are replicated in other samples. Moreover, some studies have raised the concern that effect size estimates from smaller samples, even if statistically significant, may be unreliable (Kraemer, Mintz, Noda, Tinklenberg, & Yesavage, 2006). Although it is not clear whether 689 repeated observations should be considered small for a GEE with a hurdle-negative binomial outcome, given the low variability in alcohol outcomes, this possibility should be considered, and it would be important to replicate the current findings to determine the degree to which the effect size estimates are reliable.

There are several strengths to the current study. First, we modeled our alcohol outcomes in a way that accounted for the heavily zero-inflated and skewed nature of the data and avoided combining across outcomes when doing so has been shown to produce misestimation (McGinley & Curran, 2014). Our application of multiple methods (such as testing between- and within-individual effects of negative life events, count and perceived life events, and childhood vs. concurrent ADHD) allowed a more nuanced examination of the current hypotheses. This is particularly important in light of the increasing awareness of p-hacking (Simmons, Nelson, & Simonsohn, 2011), practices that bias research studies toward presenting positive findings. As such, we intentionally presented all of our findings across all operationalizations of predictors and outcomes and used a relatively conservative approach to alpha correction with the Benjamini-Hochberg correction to provide a clear and hopefully reliable picture of how and when ADHD and negative life events are associated with alcohol use.

At the same time, several limitations warrant acknowledgment. Moreover, whereas we utilized zero-inflated hurdle models to account for the nonnormal distributions in alcohol outcomes, the alcohol items were ordinal in nature and not true counts. It may be that ignoring this may have inflated the estimates of alcohol use in the current study. Second, our reliance on self-report, particularly of life events, may have resulted in an underreporting in these events, particularly by the ADHD group. Moreover, we collapsed multiple categories of negative life events (both major and minor and life events of different sources). Although the goal here was to measure a general sense of the negative life event load, it could be that more precise findings could be obtained by a more fine-grained analysis of the effects of subcategories of negative life events. Unfortunately, there are few theoretically driven approaches to categorizing life events (but see Pillow, Barrera, & Chassin, 1998), and doing so for the current manuscript would have dramatically increased the risk of alpha inflation.

Despite these limitations, the current study adds significantly to the literature by examining negative life events with multiple different approaches over time. Specifically, these findings highlight the complexity of the negative life events–alcohol association and indicate the importance of examining both the number of life events as well as the perception of how negative these events are to the adolescent. Furthermore, future efforts focused on decreasing alcohol use the year following a negative life event may help reduce the escalation of adolescent alcohol use. Lastly, understanding how negative life events relate to alcohol use for adolescents with ADHD underscores the possibility that targeting negative life events, acute as well as chronic, may ultimately decrease risk for alcohol use disorder in this at risk population.

Allison, P. (1977). Testing for Interaction in multiple regression. American Journal of Sociology, 83, 144–153. 10.1086/226510

Aseltine, R. H., Jr., & Gore, S. L. (2000). The variable effects of stress on alcohol use from adolescence to early adulthood. Substance Use & Misuse, 35, 643–668. 10.3109/10826080009148415

Atkins, D. C., Baldwin, S., Zheng, C., Gallop, R. J., & Neighbors, C. (2013). A tutorial on count regression and zero-altered count models for longitudinal substance use data. Psychology of Addictive Behaviors, 27, 166–177.

Bagwell, C. L., Molina, B. S., Pelham, W. E., Jr., & Hoza, B. (2001). Attention-deficit hyperactivity disorder and problems in peer relations: Predictions from childhood to adolescence. Journal of the American Academy of Child and Adolescent Psychiatry, 40, 1285–1292. 10.1097/00004583-200111000-00008

Barkley, R. A. (2006). Attention-deficit hyperactivity disorder: A handbook for diagnosis and treatment (Vol. 628). New York, New York: Guilford Press Publications.

Barkley, R. A., Murphy, K. R., & Fischer, M. (2008). ADHD in adults: What the science says. New York City, NY: Guilford Press.

Beck, A. T., Steer, R. A., & Carbin, M. G. (1988). Psychometric properties of the Beck Depression Inventory: Twenty-five years of evaluation. Clinical Psychology Review, 8, 77–100. 10.1016/0272-7358(88)90050-5

Belendiuk, K. A., Pedersen, S. L., King, K. M., Pelham, W. E., & Molina, B. S. G. (2016). Change over time in adolescent and friend alcohol use: Differential associations for youth with and without childhood attention-deficit/hyperactivity disorder (ADHD). Psychology of Addictive Behaviors, 30, 29–38. 10.1037/adb0000117

Benjamini, Y., & Hochberg, Y. (1995). Controlling the false discovery rate: A practical and powerful approach to multiple testing. Journal of the Royal Statistical Society Series B. Methodological, 57, 289–300.

Bollen, K. A., & Curran, P. (2006). Latent curve models: A structural equation perspective. Hoboken, NJ: Wiley.

Cerbone, F. G., & Larison, C. L. (2000). A bibliographic essay: The relationship between stress and substance use. Substance Use & Misuse, 35, 757–786. 10.3109/10826080009148420

Chassin, L., Colder, C. R., Hussong, A. M., & Sher, K. J. (2016). Substance use and substance use disorders. In D.Cicchetti (Ed.), Developmental psychopathology: Vol. 3. Maladaptation and Psychopathology (3rd ed., pp. 833–897). Hoboken, NJ: Wiley. 10.1002/9781119125556.devpsy319

Compas, B. E., Davis, G. E., Forsythe, C. J., & Wagner, B. M. (1987). Assessment of major and daily stressful events during adolescence: The Adolescent Perceived Events Scale. Journal of Consulting and Clinical Psychology, 55, 534–541. 10.1037/0022-006X.55.4.534

Costello, E. J., & Maughan, B. (2015). Annual research review: Optimal outcomes of child and adolescent mental illness. Journal of Child Psychology and Psychiatry, and Allied Disciplines, 56, 324–341. 10.1111/jcpp.12371

Counts, C. A., Nigg, J. T., Stawicki, J. A., Rappley, M. D., & von Eye, A. (2005). Family adversity in DSM–IV ADHD combined and inattentive subtypes and associated disruptive behavior problems. Journal of the American Academy of Child and Adolescent Psychiatry, 44, 690–698. 10.1097/01.chi.0000162582.87710.66

Curran, P. J., & Bauer, D. J. (2011). The disaggregation of within-person and between-person effects in longitudinal models of change. Annual Review of Psychology, 62, 583–619. 10.1146/annurev.psych.093008.100356

Curran, P. J., Howard, A. L., Bainter, S. A., Lane, S. T., & McGinley, J. S. (2014). The separation of between-person and within-person components of individual change over time: A latent curve model with structured residuals. Journal of Consulting and Clinical Psychology, 82, 879–894. 10.1037/a0035297

Davis, G. E., & Compas, B. E. (1986). Cognitive appraisal of major and daily stressful events during adolescence: A multidimensional scaling analysis. Journal of Youth and Adolescence, 15, 377–388. 10.1007/BF02143520

Duggal, S., Malkoff-Schwartz, S., Birmaher, B., Anderson, B. P., Matty, M. K., Houck, P. R., . . .Frank, E. (2000). Assessment of life stress in adolescents: Self-report versus interview methods. Journal of the American Academy of Child and Adolescent Psychiatry, 39, 445–452. 10.1097/00004583-200004000-00013

Enders, C. K., & Tofighi, D. (2007). Centering predictor variables in cross-sectional multilevel models: A new look at an old issue. Psychological Methods, 12, 121–138. 10.1037/1082-989X.12.2.121

Evangelista, N. M., Owens, J. S., Golden, C. M., & Pelham, W. E., Jr. (2008). The positive illusory bias: Do inflated self-perceptions in children with ADHD generalize to perceptions of others?Journal of Abnormal Child Psychology, 36, 779–791. 10.1007/s10802-007-9210-8

Grant, B. F., Goldstein, R. B., Saha, T. D., Chou, S. P., Jung, J., Zhang, H., . . .Hasin, D. S. (2015). Epidemiology of DSM–5 alcohol use disorder: Results from the National Epidemiologic Survey on Alcohol and Related Conditions III. Journal of the American Medical Association Psychiatry, 72, 757–766. 10.1001/jamapsychiatry.2015.0584

Grant, K. E., Compas, B. E., Stuhlmacher, A. F., Thurm, A. E., McMahon, S. D., & Halpert, J. A. (2003). Stressors and child and adolescent psychopathology: Moving from markers to mechanisms of risk. Psychological Bulletin, 129, 447–466. 10.1037/0033-2909.129.3.447

Grant, K. E., Compas, B. E., Thurm, A. E., McMahon, S. D., & Gipson, P. Y. (2004). Stressors and child and adolescent psychopathology: Measurement issues and prospective effects. Journal of Clinical Child and Adolescent Psychology, 33, 412–425. 10.1207/s15374424jccp3302_23

Hampel, P., Manhal, S., Roos, T., & Desman, C. (2008). Interpersonal coping among boys with ADHD. Journal of Attention Disorders, 11, 427–436. 10.1177/1087054707299337

Hilbe, J. M. (2011). Negative binomial regression. Public Administration Review, 70, 1–6.

Howard, A. L., Molina, B. S. G., Swanson, J. M., Hinshaw, S. P., Belendiuk, K. A., Harty, S. C., . . .Wigal, T. (2015). Developmental progression to early adult binge drinking and marijuana use from worsening versus stable trajectories of adolescent attention deficit/hyperactivity disorder and delinquency. Addiction, 110, 784–795. 10.1111/add.12880

Hoza, B., Gerdes, A. C., Hinshaw, S. P., Arnold, L. E., Pelham, W. E., Jr., Molina, B. S. G., . . .Wigal, T. (2004). Self-perceptions of competence in children with ADHD and comparison children. Journal of Consulting and Clinical Psychology, 72, 382–391. 10.1037/0022-006X.72.3.382

Hussong, A., Bauer, D., & Chassin, L. (2008). Telescoped trajectories from alcohol initiation to disorder in children of alcoholic parents. Journal of Abnormal Psychology, 117, 63–78. 10.1037/0021-843X.117.1.63

Izard, C. E. (2002). Translating emotion theory and research into preventive interventions. Psychological Bulletin, 128, 796–824. 10.1037/0033-2909.128.5.796

Jessor, R., Donovan, J. E., & Costa, F. M. (1989). Health behavior questionnaire. Boulder, CO: Institute of Behavioral Science, University of Colorado.

Kent, K. M., Pelham, W. E., Jr., Molina, B. S. G., Sibley, M. H., Waschbusch, D. A., Yu, J., . . .Karch, K. M. (2011). The academic experience of male high school students with ADHD. Journal of Abnormal Child Psychology, 39, 451–462. 10.1007/s10802-010-9472-4

Kessler, R. C., Adler, L. A., Berglund, P., Green, J. G., McLaughlin, K. A., Fayyad, J., . . .Zaslavsky, A. M. (2014). The effects of temporally secondary co-morbid mental disorders on the associations of DSM–IV ADHD with adverse outcomes in the U.S. National Comorbidity Survey Replication Adolescent Supplement (NCS-A). Psychological Medicine, 44, 1779–1792. 10.1017/S0033291713002419

Keyes, K. M., Hatzenbuehler, M. L., Grant, B. F., & Hasin, D. S. (2012). Stress and alcohol: Epidemiologic evidence. Alcohol Research: Current Reviews, 34, 391–400.

King, K. M., & Chassin, L. (2008). Adolescent stressors, psychopathology, and young adult substance dependence: A prospective study. Journal of Studies on Alcohol and Drugs, 69, 629–638. 10.15288/jsad.2008.69.629

King, K. M., Molina, B. S. G., & Chassin, L. (2008). A state-trait model of negative life event occurrence in adolescence: Predictors of stability in the occurrence of stressors. Journal of Clinical Child and Adolescent Psychology, 37, 848–859. 10.1080/15374410802359643

King, K. M., Molina, B. S. G., & Chassin, L. (2009). Prospective relations between growth in drinking and familial stressors across adolescence. Journal of Abnormal Psychology, 118, 610–622. 10.1037/a0016315

Knopik, V. S., Heath, A. C., Jacob, T., Slutske, W. S., Bucholz, K. K., Madden, P. A. F., . . .Martin, N. G. (2006). Maternal alcohol use disorder and offspring ADHD: Disentangling genetic and environmental effects using a children-of-twins design. Psychological Medicine, 36, 1461–1471. 10.1017/S0033291706007884

Kraemer, H. C., Mintz, J., Noda, A., Tinklenberg, J., & Yesavage, J. A. (2006). Caution regarding the use of pilot studies to guide power calculations for study proposals. Archives of General Psychiatry, 63, 484–489. 10.1001/archpsyc.63.5.484

Lazarus, R. S., & Folkman, S. (1984). Coping and adaptation. In W. D.Gentry (Ed.), The Handbook of Behavioral Medicine (pp. 282–325). New York, NY: Guilford.

Lee, S. S., Humphreys, K. L., Flory, K., Liu, R., & Glass, K. (2011). Prospective association of childhood attention-deficit/hyperactivity disorder (ADHD) and substance use and abuse/dependence: A meta-analytic review. Clinical Psychology Review, 31, 328–341. 10.1016/j.cpr.2011.01.006

Marshal, M. P., Molina, B. S. G., Pelham, W. E., & Cheong, J. (2007). Attention-deficit hyperactivity disorder moderates the life stress pathway to alcohol problems in children of alcoholics. Alcoholism, Clinical and Experimental Research, 31, 564–574.

Marshal, M. P., Molina, B. S. G., & Pelham, W. E., Jr. (2003). Childhood ADHD and adolescent substance use: An examination of deviant peer group affiliation as a risk factor. Psychology of Addictive Behaviors, 17, 293–302. 10.1037/0893-164X.17.4.293

McGinley, J. S., & Curran, P. J. (2014). Validity concerns with multiplying ordinal items defined by binned counts: An application to a quantity-frequency measure of alcohol use. Methodology, 10, 108–116. 10.1027/1614-2241/a000081

McMahon, S. D., Grant, K. E., Compas, B. E., Thurm, A. E., & Ey, S. (2003). Stress and psychopathology in children and adolescents: Is there evidence of specificity?Journal of Child Psychology and Psychiatry, and Allied Disciplines, 44, 107–133. 10.1111/1469-7610.00105

Meich, R. A., Johnston, L. D., O’Malley, P. M., Bachman, J. G., & Schulenberg, J. E. (2015). Monitoring the future: National survey results on drug use, 1975–2014. Bethesda, MD: National Institutes of Health.

Molina, B. S. G., Marshal, M. P., Pelham, W. E., Jr., & Wirth, R. J. (2005). Coping skills and parent support mediate the association between childhood attention-deficit/hyperactivity disorder and adolescent cigarette use. Journal of Pediatric Psychology, 30, 345–357. 10.1093/jpepsy/jsi029

Molina, B. S. G., Pelham, W. E., Cheong, J., Marshal, M. P., Gnagy, E. M., & Curran, P. J. (2012). Childhood attention-deficit/hyperactivity disorder (ADHD) and growth in adolescent alcohol use: The roles of functional impairments, ADHD symptom persistence, and parental knowledge. Journal of Abnormal Psychology, 121, 922–935. 10.1037/a0028260

Molina, B. S. G., Pelham, W. E., Gnagy, E. M., Thompson, A. L., & Marshal, M. P. (2007). Attention-deficit/hyperactivity disorder risk for heavy drinking and alcohol use disorder is age specific. Alcoholism, Clinical and Experimental Research, 31, 643–654.

Molina, B. S. G., & Pelham, W. E., Jr. (2003). Childhood predictors of adolescent substance use in a longitudinal study of children with ADHD. Journal of Abnormal Psychology, 112, 497–507. 10.1037/0021-843X.112.3.497

Molina, B. S. G., & Pelham, W. E., Jr. (2014). Attention-deficit/hyperactivity disorder and risk of substance use disorder: Developmental considerations, potential pathways, and opportunities for research. Annual Review of Clinical Psychology, 10, 607–639. 10.1146/annurev-clinpsy-032813-153722

Mrug, S., Hoza, B., & Bukowski, W. M. (2004). Choosing or being chosen by aggressive-disruptive peers: Do they contribute to children’s externalizing and internalizing problems?Journal of Abnormal Child Psychology, 32, 53–65. 10.1023/B:JACP.0000007580.77154.69

Pelham, W. E., Jr., Gnagy, E. M., Greenslade, K. E., & Milich, R. (1992). Teacher ratings of DSM–III–R symptoms for the disruptive behavior disorders. Journal of the American Academy of Child and Adolescent Psychiatry, 31, 210–218. 10.1097/00004583-199203000-00006

Pelham, W. E., Jr., & Hoza, B. (1996). Intensive treatment: A summer treatment program for children with ADHD. In E. D.Hibbs & P. S.Jensen (Eds.), Psychosocial treatments for child and adolescent disorders: Empirically based strategies for clinical practice (pp. 311–340). Washington, DC: American Psychological Association. 10.1037/10196-013

Pillow, D. R., Barrera, M. J., & Chassin, L. (1998). Using cluster analysis to assess the effects of stressful life events: Probing the impact of parental alcoholism on child stress and substance use. Journal of Community Psychology, 26, 361–380. 10.1002/(SICI)1520-6629(199807)26:4<361::AID-JCOP5>3.0.CO;2-W

Raftery, A. E. (1995). Bayesian model selection in social research. Sociological Methodology, 25, 111–163. 10.2307/271063

Raudenbush, S. W., & Bryk, A. S. (2002). Advanced Quantitative Techniques in the Social Sciences: Vol. 1. Hierarchical linear models: Applications and data analysis methods (2nd ed.). New York City, NY: Sage.

Selzer, M. L., Vinokur, A., & van Rooijen, L. (1975). A self-administered Short Michigan Alcoholism Screening Test (SMAST). Journal of Studies on Alcohol, 36, 117–126. 10.15288/jsa.1975.36.117

Sher, K. J., Grekin, E. R., & Williams, N. A. (2005). The development of alcohol use disorders. Annual Review of Clinical Psychology, 1, 493–523. 10.1146/annurev.clinpsy.1.102803.144107

Sibley, M. H., Graziano, P. A., Kuriyan, A. B., Coxe, S., Pelham, W. E., Rodriguez, L., . . .Ward, A. (2016). Parent-teen behavior therapy + motivational interviewing for adolescents with ADHD. Journal of Consulting and Clinical Psychology, 84, 699–712. 10.1037/ccp0000106

Sibley, M. H., Pelham, W. E., Jr., Molina, B. S. G., Gnagy, E. M., Waschbusch, D. A., Garefino, A. C., . . .Karch, K. M. (2012). Diagnosing ADHD in adolescence. Journal of Consulting and Clinical Psychology, 80, 139–150. 10.1037/a0026577

Simmons, J. P., Nelson, L. D., & Simonsohn, U. (2011). False-positive psychology: Undisclosed flexibility in data collection and analysis allows presenting anything as significant. Psychological Science, 22, 1359–1366. 10.1177/0956797611417632

Spitzer, R., Williams, J., Gibbon, M., & First, M. (1990). User’s guide for the structured clinical interview for DSM–III–R: SCID. Washington, DC: American Psychiatric Association.

The Substance Abuse and Mental Health Services Administration. (1992). National household survey on drug abuse. Rockville, MD: U.S. Government Printing Office.

Vatcheva, K. P., Lee, M., McCormick, J. B., & Rahbar, M. H. (2015). The effect of ignoring statistical interactions in regression analyses conducted in epidemiologic studies: An example with survival analysis using Cox proportional hazards regression model. Epidemiology, 6, 216. 10.4172/2161-1165.1000216

Wagner, B. M., & Compas, B. E. (1990). Gender, instrumentality, and expressivity: Moderators of the relation between stress and psychological symptoms during adolescence. American Journal of Community Psychology, 18, 383–406. 10.1007/BF00938114

Walther, C. A. P., Cheong, J., Molina, B. S. G., Pelham, W. E., Jr., Wymbs, B. T., Belendiuk, K. A., & Pedersen, S. L. (2012). Substance use and delinquency among adolescents with childhood ADHD: The protective role of parenting. Psychology of Addictive Behaviors, 26, 585–598. 10.1037/a0026818

Wills, T. A., Sandy, J. M., & Yaeger, A. M. (2002). Moderators of the relation between substance use level and problems: Test of a self-regulation model in middle adolescence. Journal of Abnormal Psychology, 111, 3–21. 10.1037/0021-843X.111.1.3

Wills, T. A., Sandy, J. M., Yaeger, A. M., Cleary, S. D., & Shinar, O. (2001). Coping dimensions, life stress, and adolescent substance use: A latent growth analysis. Journal of Abnormal Psychology, 110, 309–323. 10.1037/0021-843X.110.2.309

Wymbs, B. T., Pelham, W. E., Jr., Molina, B. S. G., Gnagy, E. M., Wilson, T. K., & Greenhouse, J. B. (2008). Rate and predictors of divorce among parents of youths with ADHD. Journal of Consulting and Clinical Psychology, 76, 735–744. 10.1037/a0012719

Zeger, S. L., Liang, K. Y., & Albert, P. S. (1988). Models for longitudinal data: A generalized estimating equation approach. Biometrics, 44, 1049–1060. 10.2307/2531734

Zeileis, A., Kleiber, C., & Jackman, S. (2008). Regression models for count data in R. Journal of Statistical Software, 27, 1–25. 10.18637/jss.v027.i08

Submitted: January 28, 2017 Revised: May 11, 2017 Accepted: May 12, 2017

Childhood Cognitive Measures as Predictors of Alcohol Use and Problems by Mid-Adulthood in a Non-Western Cohort / BRIEF REPORT

Acknowledgement: Supported by the National Institutes of Health (Grants K08 AA14265, R01 AA10207, and R01 AA18179), the Mauritian Ministry of Health, the Medical Research Council, and the Wellcome Trust. We thank the staff of the Joint Child Health Project for their assistance with data collection and management, the Joint Child Health Project participants for their lifelong participation in this project, and John L. Horn, PhD, for his mentorship on this study.

Developmental models of the etiology of alcohol problems have proposed several pathways for how childhood cognitive deficits relate to later alcohol problems. Deviance proneness models purport that underlying behavioral dysregulation manifests in part as cognitive deficits and poor academic performance in childhood and later as alcohol problems in adulthood (see Gorenstein & Newman, 1980; Sher, 1991; Zucker, Chermack, & Curran, 2000). Alternative models have proposed more direct paths—that high intelligence and academic success can lead to both increased likelihood of being a lifetime drinker by placing individuals in heavier drinking environments (e.g., college) as well as decreased likelihood of developing alcohol problems via better opportunities in adulthood that may buffer against problems (see Johnson, Hicks, McGue, & Iacono, 2009). Conversely, pressure to maintain academic achievement may result in distress (e.g., Stoeber & Rambow, 2007), which in turn can lead to alcohol use and problems (Crum et al., 2006; Schulenberg, Bachman, O’Malley, & Johnston, 1994). It is also possible that multiple processes are involved, with both behavioral undercontrol and contextual factors contributing to the link between early cognitive deficits and later alcohol involvement.

General population sample studies have demonstrated that relatively higher IQs are typically found among low-to-moderate drinkers compared with abstainers and heavy drinkers (see Anstey, Windsor, Rodgers, Jorm, & Christensen, 2005; Müller et al., 2013, for reviews). In longitudinal studies, higher childhood IQ has been positively associated with alcohol use and higher consumption levels in early and later adulthood (Johnson, Hicks, McGue, & Iacono, 2009; Kanazawa & Hellberg, 2010). For example, in a U.S. national sample of young adults assessed over a 5-year interval, higher verbal IQ predicted increased risk for subsequent drinking and decreased risk for problems among drinkers even after covarying for socioeconomic status (Windle & Blane, 1989). This is consistent with a prospective study of a general Scottish sample that found higher verbal IQ at 11 years was associated with alcohol problems 40 years later after covarying for socioeconomic position (Batty, Deary, & Macintyre, 2006). Taken together, these studies suggest that higher IQ, and in particular verbal IQ, is predictive of increased likelihood of being a lifetime drinker and decreased likelihood of alcohol-related problems later in life, and that these associations are not accounted for by sociodemographic correlates of higher verbal abilities.

Academic achievement is also a potential childhood cognitive predictor of subsequent alcohol involvement. Support for poor academic achievement associated with subsequent alcohol use and problems also has been found in longitudinal studies and general samples with varying ranges of follow-up (see, e.g., Duncan, Duncan, Biglan, & Ary, 1998; Hawkins, Catalano, & Miller, 1992; Schulenberg et al., 1994). For example, Hayatbakhsh, Najman, Bor, Clavarino, and Alati (2011) demonstrated that poorer school performance at age 14 years predicted alcohol problems 21 years later in a general sample of Australian students, although already at age 14 over 60% of the sample indicated drinking in the past week. Studies that examine academic achievement in younger samples are needed to help clarify if poor academic achievement is a predictor of subsequent alcohol involvement or just a consequence of early alcohol use.

In this study, we present longitudinal data from a birth cohort sample that was assessed at age 11 years on measures of cognitive and academic ability (prior to the typical age of onset of alcohol use) and then approximately 25 years later for lifetime alcohol use and alcohol use disorder (AUD) symptoms. The sample was from the island of Mauritius (a middle-income eastern African nation), allowing for the examination of these relationships in a non-Western society that values education and academic performance (the population has an 89% literacy rate and public primary and secondary education are free; Central Intelligence Agency, 2013; Southern and Eastern Africa Consortium for Monitoring Educational Quality [SACMEQ], 2012), but where childhood cognitive performance is not linked to heavy drinking environments as it often is in Western societies (e.g., college; Slutske et al., 2004). However, as in Western societies, both intelligence and school success in Mauritius may enable individuals to obtain financial and personal resources that increase the likelihood and opportunities for social drinking, even if buffering against risk for alcohol problems (see Johnson et al., 2009; Müller et al., 2013). Such contextual factors may affect relationships between childhood cognitive performance and subsequent alcohol involvement; thus, examining these associations in novel societies such as Mauritius will help determine the generalizability of the developmental models that have been generated using data primarily from Western societies (see Luczak et al., 2014). Given our prior findings with this sample that found gender and an index of psychosocial adversity (based on familial, housing, and environmental variables) were associated with IQ (Liu, Raine, Venables, & Mednick, 2004; Lynn, Raine, Venables, Mednick, & Irwing, 2005), and that being Muslim was protective for lifetime drinking but not for alcohol problems among drinkers (Luczak et al., 2014) we recognized the importance of including gender, psychosocial adversity, and Muslim religion when examining the link between cognition and alcohol use in this sample.

Data were from the Joint Child Health Project (JCHP), a longitudinal study in Mauritius that has followed a birth cohort of 1,795 children since 1972 when they were 3 years old (see Raine, Liu, Venables, Mednick, & Dalais, 2010). The original sample was 51% male and, similar to the population of the island, 69% were of Indian heritage, 26% Creole (admixture of primarily African descent), and 6% other (primarily of Chinese and French heritage); 17% were Muslim.

Childhood Data Collection Phase

At age 11 years, participants were assessed on cognitive ability and home environment (see each scale for n values, which differ across scales because a cyclone brought this testing phase to an early end). The 11-year-old sample did not differ significantly from the 3-year-old sample on gender, ethnicity, or psychosocial adversity (see Raine, Reynolds, Venables, Mednick, & Farrington, 1998).

All instruments were administered to children individually by trained research staff. The official language of Mauritius is English, and schooling is conducted primarily in English; but because the common spoken language on the island is Kreol, instructions were given in Kreol. Seven subtests of the Wechsler Intelligence Scale for Children (WISC; Wechsler, 1949) were administered to 1,258 children. These subtests were used to create estimates of Performance IQ (PIQ; Picture Completion, Block Design, Object Assembly, Coding, and Mazes) and Verbal IQ (VIQ; Similarities and Digit Span). Scores were standardized and normalized within the sample (see Lynn et al., 2005 and Raine, Yaralian, Reynolds, Venables, & Mednick, 2002, for details).

The Trail Making Test (TMT; U.S. Army, 1944) was administered to 1,157 of the children. The two components of this task (Parts A [TMT-A] and B [TMT-B]) assess visuomotor tracking, motor speed, and attention, and TMT-B also contains a working memory component requiring mental flexibility (Lezak, 1995; Reitan, 1958). The difference score for TMT-B versus TMT-A (TMT B-A), an indicator of complex divided attention and sequencing (Strauss, Sherman, & Spreen, 2006), was calculated and corrected for age (M = 11.1 years, SD = 0.70) by residualization (see Raine, Reynolds, Venables, & Mednick, 2002).

At the end of primary school (sixth year of school), students take the Certificate of Primary Education (CPE) achievement examination that covers a range of academic topics (English, French, Mathematics, Environmental Studies; see SACMEQ, 2012). Unweighted scores on the CPE range from 0−20 and represent an overall index of academic achievement. CPE scores were obtained from official records for 1,415 of the sample.

Midadulthood Data Collection Phase

In mid–adulthood (M = 36.9 years, SD = 1.39), all available participants (n = 1,209 [67%]) were assessed for lifetime alcohol use and problems (other 9% abroad, 4% refused, 2% deceased, and 18% unable to contact). Written informed consent was obtained, and the research was approved by the University of Southern California Institutional Review Board. The sample assessed in adulthood did not differ significantly from the 3-year-old sample on ethnicity or psychosocial adversity, but was more likely to be male (55% vs. 51%; see Luczak et al., 2014).

Trained research staff interviewed participants about their drinking histories in Kreol. Lifetime drinkers were defined as those who had ever consumed at least one standard drink (about 14 g of alcohol), and the age when the first standard drink was consumed was obtained. Lifetime drinkers were assessed for lifetime Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM–IV) AUD symptoms (American Psychiatric Association, 1994) using the Structured Clinical Interview for DSM–IV Diagnosis (Spitzer, Gibbon, & Williams, 1997. Only 15 (1.2%) participants reported they had consumed a standard drink and none (0%) endorsed having an AUD symptom as of age 11 years.

Final Analytic Sample

We removed four participants who were developmentally delayed, resulting in a final analytic sample of 1,107 with childhood cognitive and adult alcohol data (see Table 1). This sample was 72% Indian, 21% Creole, and 7% other; 55% were male and 22% Muslim. Scores on the cognitive measures and psychosocial adversity did not differ significantly from the full sample assessed at age 11 years. Lifetime drinker (66%) and AUD (16%) prevalence were similar to those previously reported for the full sample assessed in midadulthood (67% and 16%, respectively; Luczak et al., 2014). In those who endorsed an AUD symptom (n = 205), mean symptom count was 3.7 (SD = 2.46; range: 1–11).

Intercorrelations Among Cognitive Measures and Demographic Covariates

Data Analyses

We ran logistic regression models to examine cognitive predictors and covariates of being a lifetime drinker. We used zero-inflated negative binomial (ZINB) regression models to examine cognitive predictors and demographic covariates of AUD symptom count. ZINB is a two-part parametric mixture model for count data that have a large proportion of zero values and a highly skewed distribution of nonzero values, as is typically found for AUD symptoms in general populations samples (see Pardini, White, & Stouthamer-Loeber, 2007). ZINB models are also appropriate when there is heteroscedasticity in the count, which may occur if one covariate group (e.g., males, Muslims) produces different counts than another (Neelon & O’Malley, 2014).

We first modeled each cognitive predictor alone, then with demographic covariates (i.e., gender, childhood psychosocial adversity, and Muslim religion). All predictor variables were normally distributed (skew < |.90|, all kurtosis < |1.76|; note that we divided IQ variables by 10 to put all predictors on similar scales to yield more interpretable betas). We ran a final set of models that simultaneously entered the cognitive measures that were significant individual predictors of AUD symptoms to examine these cognitive variables in concert. Models with significant quadratic or interaction terms (created through cross-multiplication based on centered predictors) are presented only when one of these terms was significant.

Table 1 shows basic associations among the predictors in our models. Consistent with our prior publications (Liu et al., 2004; Lynn et al., 2005; Yarnell et al., 2013), being male correlated with higher PIQ, childhood psychosocial adversity correlated with poorer cognitive performance, and the three demographic variables did not correlate with one another.

Predictors of Lifetime Drinking

Logistic regression models found that those performing better on the TMT B-A were more likely to be lifetime drinkers, even after the addition of covariates in the model (b = −2.48, p = .017). A significant association of PIQ (b = 0.14, p = .005) with lifetime drinking in the univariate model was reduced to nonsignificant with the addition of covariates (p = .73). Neither VIQ nor CPE scores were significantly associated with lifetime drinking, with or without covariates.

Predictors of AUD Symptoms

Table 2 shows results of the symptom count portion of the ZINB models for each cognitive predictor alone and with the three demographic covariates.

Childhood Cognitive Predictors of Lifetime Alcohol Use Disorder Symptom Count With and Without Covarying for Gender, Childhood Psychosocial Adversity, and Muslim Religion

Intellectual ability

In univariate models, lower scores on each IQ scale predicted AUD symptoms (ps < .04), but only VIQ remained a significant predictor with covariates (PIQ reduced to p = .09). We found one significant VIQ × Male interaction (b = 0.05, p = .017), which we probed by rerunning the model separately for each gender. Lower VIQ was a stronger predictor of AUD symptoms for females (b = −0.53, p < .001) than for males (b = −0.08, p = .07).

Trail Making Test

Models including nonlinear terms revealed a significant negative effect of (TMT B-A)² on AUD symptoms (b = 0.07, p < .05). We probed this effect by reestimating the model with the TMT B-A distribution trichotomized into three groups (fast = < −1 SD; midrange = between −1 and +1 SD; slow = > 1 SD). The relationship of TMT B-A with AUD symptom count differed in both direction and magnitude across the three groups (fast b = −1.13, p = .79; midrange b = 0.94, p = .55; and slow b = 2.61, p = .05). Only for those in the slow range was there indication of TMT B-A being predictive of AUD symptoms.

Academic achievement

Lower CPE scores were predictive of higher AUD symptoms, with (p = .029) and without (p = .012) covariates included in the model.

Multiple cognitive predictors

Lastly, we entered the two cognitive variables that were significant in linear models, VIQ and CPE scores, simultaneously in ZINB models with and without covariates. When entered together, neither predicted AUD symptoms (ps > .24), indicating the portion of each of these associated with AUD symptoms may be shared.

This study examined cognitive abilities and academic achievement in a general cohort, non-Western sample of 11-year-old youth as predictors of alcohol involvement over approximately the next 25 years. Lifetime drinking was predicted only by better childhood performance on the TMT B-A once demographic covariates were taken into account. Lower verbal abilities and academic achievement were linearly associated with subsequent alcohol problems, whereas the relationship between poorer performance on the TMT B-A and alcohol problems only emerged at the lower end of the performance range. All of these associations were found prospectively in a general sample of youth tested prior to the typical onset of drinking in this society (99% of the sample had not consumed a full drink), indicating these associations existed prior to alcohol use and were not merely consequences of consumption.

Being a lifetime drinker was not strongly linked to childhood cognitive measures in this Mauritian sample after accounting for demographic covariates. Studies of twin samples have found that initiation of alcohol use is more strongly influenced by environmental factors than genetic factors (e.g., Heath, Meyer, Eaves, & Martin, 1991; Rhee et al., 2003), and thus factors associated with lifetime drinking status may vary more across cultures. One environmental explanation that has been proposed for the higher levels of drinking seen among those with higher intellectual ability and achievement in Western societies is that this increases the opportunity for alcohol use (e.g., Johnson et al., 2009; Slutske et al., 2004). On Mauritius, however, higher academic achievement does not typically place young adults in a more risky drinking environment, given that higher education is not linked to moving away from one’s family of origin, nor does high academic achievement necessarily allow one to pursue higher education. Thus, our findings are consistent with the idea that exposure to drinking environments contributes to the link between childhood cognitive performance and lifetime drinking. Additional examination of differences in drinking norms, contexts, and accessibility across societies may further elucidate unique environmental components of this relationship.

On the other hand, both childhood verbal intelligence and academic achievement were predictive of lifetime alcohol-related problems, even after accounting for gender, childhood psychosocial adversity, and Muslim religion in this Mauritian sample. These results are consistent with other prospective cohort studies in Western societies (e.g., Batty et al., 2006; Windle & Blane, 1989), providing further evidence that poor cognitive performance in childhood is a robust predictor of later alcohol problems across cultures.

In our sample, verbal intelligence was a stronger predictor of lifetime alcohol problems in females than in males. The specificity of the relationship of lower verbal abilities with alcohol problems in our female sample suggests the possibility of a pathway to alcohol problems that is distinct from the deviance proneness pathway (which is more associated with behavioral undercontrol in males; Sher, 1991) that may operate more through other factors such as social skills and judgment (see Maggs, Patrick, & Feinstein, 2008; Windle & Blane, 1989). Further examination of gender differences in models that include social, peer, and situational factors may help explain the association between verbal abilities and alcohol-related problems.

Poor academic achievement in childhood was found to be a significant predictor of lifetime alcohol problems in both males and females. This relationship remained significant after covarying for psychosocial adversity, even though previous reports with this sample have shown psychosocial adversity to be associated with lower IQ scores (Liu et al., 2004). In Mauritius, primary education is mandated, and the children in our cohort had relatively uniform educational opportunities up through age 11 years, regardless of psychosocial adversity. A mediational pathway cannot be established with these data, but our findings indicate that poor academic achievement by the end of primary school is a risk factor for subsequent alcohol problems that is not simply accounted for by concurrent psychosocial adversity.

The measure of TMT B-A exhibited a curvilinear relationship with lifetime alcohol problems. More severe deficits served as a risk factor, but stronger performance did not add any protection against developing alcohol problems. Our findings suggest that, within a general sample, the link between poor functioning on the TMT B-A in childhood and subsequent alcohol problems emerges only for those in the lowest performance range. The deviance proneness model hypothesizes that underlying behavioral dysregulation manifests in childhood as attention and cognitive deficits; the TMT B-A difference score, a measure of complex divided attention and sequencing (Strauss et al., 2006), may be assessing abilities that are indicators of this dysregulation construct. Future research examining general samples that include individuals on the tail ends of performance will improve our understanding of the specificity of various cognitive measures as precursors to alcohol problems as well as their relationships with other measures of behavioral dysregulation.

Our findings should be interpreted along with the limitations of the study, including the use of a limited number of measures to assess cognitive performance in a culture for which they were not designed, attrition as is the case in any longitudinal study, and the inability to speak to mechanisms for how cognitive deficits in childhood lead to alcohol problems by adulthood. Despite these limitations, this study provides evidence that childhood cognitive performance is not strong a predictor of lifetime alcohol use in this society, indicating environmental specificity of previously found relationships in Western societies, but that childhood cognitive deficits are risk factors for subsequent alcohol problems in this novel east African cultural context, providing further evidence of the generalizability of this relationship across societies.

American Psychiatric Association. (1994). Diagnostic and statistical manual of mental disorders, fourth edition (DSM–IV). Washington, DC: Author.

Anstey, K. J., Windsor, T. D., Rodgers, B., Jorm, A. F., & Christensen, H. (2005). Lower cognitive test scores observed in alcohol abstainers are associated with demographic, personality, and biological factors: The PATH Through Life Project. Addiction, 100, 1291–1301. 10.1111/j.1360-0443.2005.01159.x

Batty, G. D., Deary, I. J., & Macintyre, S. (2006). Childhood IQ and life course socioeconomic position in relation to alcohol induced hangovers in adulthood: The Aberdeen children of the 1950s study. Journal of Epidemiology and Community Health, 60, 872–874. 10.1136/jech.2005.045039

Central Intelligence Agency. (2013). The world factbook. Retrieved from https://www-cia-gov.offcampus.lib.washington.edu/library/publications/the-world-factbook/index.html

Crum, R. M., Juon, H.-S., Green, K. M., Robertson, J., Fothergill, K., & Ensminger, M. (2006). Educational achievement and early school behavior as predictors of alcohol-use disorders: 35-year follow-up of the Woodlawn Study. Journal of Studies on Alcohol, 67, 75–85.

Duncan, S. C., Duncan, T. E., Biglan, A., & Ary, D. (1998). Contributions of the social context to the development of adolescent substance use: A multivariate latent growth modeling approach. Drug and Alcohol Dependence, 50, 57–71. 10.1016/S0376-8716(98)00006-4

Gorenstein, E. E., & Newman, J. P. (1980). Disinhibitory psychopathology: A new perspective and a model for research. Psychological Review, 87, 301–315. 10.1037/0033-295X.87.3.301

Hawkins, J. D., Catalano, R. F., & Miller, J. Y. (1992). Risk and protective factors for alcohol and other drug problems in adolescence and early adulthood: Implications for substance abuse prevention. Psychological Bulletin, 112, 64–105. 10.1037/0033-2909.112.1.64

Hayatbakhsh, M. R., Najman, J. M., Bor, W., Clavarino, A., & Alati, R. (2011). School performance and alcohol use problems in early adulthood: A longitudinal study. Alcohol, 45, 701–709. 10.1016/j.alcohol.2010.10.009

Heath, A. C., Meyer, J., Eaves, L. J., & Martin, N. G. (1991). The inheritance of alcohol consumption patterns in a general population twin sample: I. Multidimensional scaling of quantity/frequency data. Journal of Studies on Alcohol, 52, 345–352.

Johnson, W., Hicks, B. M., McGue, M., & Iacono, W. G. (2009). How intelligence and education contribute to substance use: Hints from the Minnesota Twin Family Study. Intelligence, 37, 613–624. 10.1016/j.intell.2008.12.003

Kanazawa, S., & Hellberg, J. E. E. U. (2010). Intelligence and substance use. Review of General Psychology, 14, 382–396. 10.1037/a0021526

Lezak, M. D. (1995). Neuropsychological assessment (3rd ed.). New York, NY: Oxford University Press.

Liu, J., Raine, A., Venables, P. H., & Mednick, S. A. (2004). Malnutrition at age 3 years and externalizing behavior problems at ages 8, 11, and 17 years. The American Journal of Psychiatry, 161, 2005–2013. http://www-ncbi-nlm-nih-gov.offcampus.lib.washington.edu/pmc/articles/PMC1570126/pdf/nihms-12067.pdf

Luczak, S. E., Prescott, C. A., Dalais, C., Raine, A., Venables, P. H., & Mednick, S. A. (2014). Religious factors associated with alcohol involvement: Results from the Mauritian Joint Child Health Project. Drug and Alcohol Dependence, 135, 37–44. 10.1016/j.drugalcdep.2013.10.028

Lynn, R. T., Raine, A., Venables, P. H., Mednick, S. A., & Irwing, P. (2005). Sex differences on the WISC-R in Mauritius. Intelligence, 33, 527–533. 10.1016/j.intell.2005.05.001

Maggs, J. L., Patrick, M. E., & Feinstein, L. (2008). Childhood and adolescent predictors of alcohol use and problems in adolescence and adulthood in the National Child Development Study. Addiction, 103(Supp. 1), 7–22. 10.1111/j.1360-0443.2008.02173.x

Müller, M., Kowalewski, R., Metzler, S., Stettbacher, A., Rössler, W., & Vetter, S. (2013). Associations between IQ and alcohol consumption in a population of young males: A large database analysis. Social Psychiatry and Psychiatric Epidemiology, 48, 1993–2005. 10.1007/s00127-013-0666-2

Neelon, B., & O’Malley, A. J. (2014). Two-part models for zero-inflated and semicontinuous data. In B.Sobolev & C.Gatsonis (Eds.), Handbook of health services research. New York, NY: Springer. Retrieved from http://people.duke.edu/?neelo003/Working_Papers/tpchapter.pdf

Pardini, D., White, H. R., & Stouthamer-Loeber, M. (2007). Early adolescent psychopathology as a predictor of alcohol use disorders by young adulthood. Drug and Alcohol Dependence, 88(Supp. 1), S38–S49. 10.1016/j.drugalcdep.2006.12.014

Raine, A., Liu, J., Venables, P. H., Mednick, S. A., & Dalais, C. (2010). Cohort profile: The Mauritius Child Health Project. International Journal of Epidemiology, 39, 1441–1451. 10.1093/ije/dyp341

Raine, A., Reynolds, C., Venables, P. H., & Mednick, S. A. (2002). Stimulation seeking and intelligence: A prospective longitudinal study. Journal of Personality and Social Psychology, 82, 663–674. 10.1037/0022-3514.82.4.663

Raine, A., Reynolds, C., Venables, P. H., Mednick, S. A., & Farrington, D. P. (1998). Fearlessness, stimulation-seeking, and large body size at age 3 years as early predispositions to childhood aggression at age 11 years. Archives of General Psychiatry, 55, 745–751. 10.1001/archpsyc.55.8.745

Raine, A., Yaralian, P. S., Reynolds, C., Venables, P. H., & Mednick, S. A. (2002). Spatial but not verbal cognitive deficits at age 3 years in persistently antisocial individuals. Development and Psychopathology, 14, 25–44. 10.1017/S0954579402001025

Reitan, R. M. (1958). Validity of the Trail Making Test as an indicator of organic brain damage. Perceptual and Motor Skills, 8, 271–276. 10.2466/pms.1958.8.3.271

Rhee, S. H., Hewitt, J. K., Young, S. E., Corley, R. P., Crowley, T. J., & Stallings, M. C. (2003). Genetic and environmental influences on substance initiation, use, and problem use in adolescents. Archives of General Psychiatry, 60, 1256–1264. 10.1001/archpsyc.60.12.1256

Schulenberg, J., Bachman, J. G., O’Malley, P. M., & Johnston, L. D. (1994). High school educational success and subsequent substance use: A panel analysis following adolescents into young adulthood. Journal of Health and Social Behavior, 35, 45–62. 10.2307/2137334

Sher, K. J. (1991). Children of alcoholics: A critical appraisal of theory and research. Chicago, IL: University of Chicago Press.

Slutske, W. S., Hunt-Carter, E. E., Nabors-Oberg, R. E., Sher, K. J., Bucholz, K. K., Madden, P. A., . . .Heath, A. C. (2004). Do college students drink more than their non-college-attending peers? Evidence from a population-based longitudinal female twin study. Journal of Abnormal Psychology, 113, 530–540. 10.1037/0021-843X.113.4.530

Southern and Eastern Africa Consortium for Monitoring Educational Quality. (2012). Mauritius education fact sheet. Retrieved from http://www.sacmeq.org/education-mauritius.htm

Spitzer, M. B., Gibbon, M., & Williams, J. (1997). Structured Clinical Interview for DSM–IV Axis I disorders. New York, NY: New York State Psychiatric Institute, Biometrics Research Department.

Stoeber, J., & Rambow, A. (2007). Perfectionism in adolescent school students: Relations with motivation, achievement, and well-being. Personality and Individual Differences, 42, 1379–1389. 10.1016/j.paid.2006.10.015

Strauss, E., Sherman, E., & Spreen, O. (2006). A compendium of neuropsychological tests (3rd ed.). New York, NY: Oxford University Press.

U.S. Army. (1944). Trail Making Test. Washington, DC: Adjutant General’s Office, War Department.

Venables, P. H., Dalais, J. C., Mitchell, D. A., Mednick, S. A., & Schlusinger, F. (1983). Outcome at age nine of psychophysiological selection at age three for risk of schizophrenia: A Mauritian study. British Journal of Developmental Psychology, 1, 21–30. http://onlinelibrary.wiley.com.offcampus.lib.washington.edu/doi/10.1111/j.2044-835X.1983.tb00541.x/pdf

Wechsler, D. (1949). Manual for the Wechsler Intelligence Scale for Children. New York, NY: Psychological Corporation.

Windle, M., & Blane, H. T. (1989). Cognitive ability and drinking behavior in a national sample of young adults. Alcoholism: Clinical and Experimental Research, 13, 43–48. 10.1111/j.1530-0277.1989.tb00282.x

Yarnell, L. M., Sargeant, M. N., Prescott, C. A., Tilley, J. L., Farver, J. A. M., Mednick, S. A., . . .Luczak, S. E. (2013). Measurement invariance of internalizing and externalizing behavioral syndrome factors in a non-Western sample. Assessment, 20, 642–655. 10.1177/1073191113498114

Zucker, R. A., Chermack, S. T., & Curran, G. M. (2000). Alcoholism: A lifespan perspective on etiology and course. In A. J.Sameroff, M.Lewis, & S. M.Miller (Eds.), Handbook of developmental psychopathology (2nd ed., pp. 569–587). New York, NY: Plenum Press. 10.1007/978-1-4615-4163-9_30

Submitted: July 10, 2014 Revised: September 30, 2014 Accepted: October 6, 2014

Childhood Family Characteristics and Prescription Drug Misuse in a National Sample of Latino Adults / BRIEF REPORT

Acknowledgement: A poster using the broader National Epidemiologic Survey on Alcohol and Related Conditions data looking at associations between parental divorce and death and prescription drug misuse was presented at the American Psychological Association in 2010. The results of the current study focusing on the Latino subsample have not been previously disseminated.

Note: John M. Roll served as the action editor for this article.

The Centers for Disease Control and Prevention ([CDC]; 2016) report that overdoses from prescription drugs resulting in death increased rapidly from 1999 to 2014. The CDC analyzed the morbidity underlying these overdoses and found that number of emergency room visits for nonmedical use of opioid analgesics increased by 111% between 2004 and 2008 (Cai, Crane, Poneleit, & Paulozzi, 2010). Emergency room visits related to nonmedical benzodiazepine use increased 89% from 2004 to 2008. The misuse of prescription drugs is a public health problem that permeates the range of communities from urban cities to rural communities (Substance Abuse and Mental Health Services Administration, Center for Behavioral Health Statistics and Quality, 2013). Data from the National Survey on Drug Use and Health (Substance Abuse and Mental Health Services Administration [SAMHSA], 2015) indicate that approximately 6.5 million (2.5%) Americans use psychotherapeutic drugs for nonmedical use each month.

Research investigating prescription drug misuse among Latino populations is sparse and needed. Latinos represent a large and growing population in the United States (Stepler & Brown, 2016). It has been noted that because the population is young, Latinos are at increased risk for substance use (Volkow, 2006). Among Latino eighth graders, the rates of use for various prescription drugs such as hydrocodone are on par with or exceed use of these substances by non-Latino, White youth (Johnston, O’Malley, Miech, Bachman, & Schulenberg, 2016). The rates of use fall by 12th grade, but this may be due to high rates of dropout and school exclusion by minority youth. In adulthood and older adulthood, the rates of prescription drug misuse among Latinos is also concerning (SAMHSA, 2015). For example, a recent study showed that, among adults age 65 and older, Latinos having a greater odds of misusing prescription drugs than non-Latino Whites or African Americans (Moore et al., 2009).

The empirical literature addressing correlates of substance use among Latinos has often focused on important cultural and contextual factors. Familismo, which reflects the cultural saliency of family for Latinos, remains a powerful cultural value for Latinos (Santiago-Rivera, Arredondo, & Gallardo-Cooper, 2002). Thus, prevention and intervention efforts with Latinos are often family focused (e.g., Familias Unidas; Coatsworth, Pantin, & Szapocznik, 2002). Family factors have been related to substance use among Latino adolescents and emerging adults (Knight et al., 2010). There are a number of studies linking illicit substance use, including nonprescribed use of prescription drugs, to parental alcoholism and parental divorce in predominantly White, non-Latino samples (e.g., Waldron, Grant, et al., 2014, Waldron, Vaughan, et al., 2014). However, little is known about parental alcoholism and parental divorce before the age of 18 in Latino families and their relationship to prescription drug misuse, specifically opiates, tranquilizers, and benzodiazepines. To our knowledge, there are no published studies of parental death before the age of 18 and risk of prescription drug misuse, regardless of racial/ethnic background.

Research testing associations between family characteristics and prescription drug misuse among Latinos has had some growth, but has largely focused on adolescents. For example, Ford and Rigg (2015) found that stronger bonds with parents reduced the odds of prescription pain reliever misuse among Latino adolescents. In another study, perceived parental disapproval of marijuana and alcohol use decreased the odds of prescription drug misuse among Latinos (Conn & Marks, 2014). Two studies report associations with family structure, including parental separation or divorce. Using Monitoring the Future data, Harrell and Broman (2009) did not find an association between family structure (e.g., single parenthood and stepfamily vs. two-parent household) and prescription drug misuse among Latino adolescents. Barrett and Turner (2006) found that less problematic adolescent substance use was associated with intact mother-father families than single-parent families. However, those single-parent families that had at least one additional relative living in the household did not show the increased risk for substance use among adolescents. While this study had large subsample of Latino youth, the authors did not assess these specific relationships stratified by racial or ethnic groups.

Parental alcoholism has not been studied in relation to prescription drug misuse among Latino samples. Such an omission is surprising given long-documented research on children of alcoholics and their increased risk for early and problem involvement across a range of substance classes (Sher, 1991; Windle & Searles, 1990), including prescription drugs (Tucker et al., 2015). Information on parental alcoholism and both parental divorce and parental death during childhood is important to understanding the associations between family context and prescription drug misuse in an understudied sample of Latinos. In addition, the current study will also include nativity as an important covariate for Latino populations. Previous research has indicated that Latinos born outside of the United States have lower rates of substance use disorders (Alegria, Canino, Stinson, & Grant, 2006). The aim of the current study is to test association between family characteristics prior to age 18 and prescription drug misuse in a national sample of Latino adults. Important demographic covariates such as biological sex and nativity will be included in models testing these associations.

This study is a secondary data analysis of data from wave one of the NESARC. The NESARC is a representative sample of noninstitutionalized adults in the United States and District of Columbia (Chen et al., 2006; Grant & Dawson, 2006). Participants’ data are gathered through in-home computer assisted interviews. The primary goal of NESARC is to ascertain the prevalence of alcohol use disorders and related problems. Data are collected on a broad spectrum of substance use behaviors including use of prescription drugs without a prescription. Review by the institutional review board determined that the research was exempt from review due to the publically available and deidentified nature of the data.

Participants

Participants for this secondary data analysis were self-identified Latino adults who participated in wave one of the NESARC (n = 8,308). The sample was nearly evenly split between men and women (49.1% female) and over half of the sample was born outside of the United States (57.3%). With respect to prescription drug misuse, 4.9% of the sample reported lifetime prescription drug misuse and 1.4% reported lifetime prescription drug use disorder. More detailed demographic information can be found in Table 1.

Frequencies

Measures

Demographic variables

Demographic variables included biological sex, marital status, income, and nativity. Biological sex (female = 0; male = 1), marital status (not married = 0; married = 1), and nativity (0 = born outside of the U.S.; 1 = born in the U.S.) were all dummy coded. Personal income was a four-level categorical variable representing, 0 = $0–$19,999, 1 = $20,000–$49,999, 2 = $50,000–$79,999, and 3 = $80,000 and up.

Childhood family characteristics

Participants were asked whether they had experienced a number of family transitions prior to the age of 18. For the current study, we tested whether participants experienced divorce of a parent (0 = no, 1 = yes) and whether the participant experienced the death of a parent (0 = no, 1 = yes).

Parental alcoholism

Two items assessed maternal and paternal alcoholism. Participants were asked, “Has your blood or natural <parent> been an alcoholic or problem drinker at any time in <his/her> life?”. These two items were combined and coded to indicate whether or not the participant’s parent had parental a history of alcoholism (0 = no, 1 = yes).

Substance use disorders

Alcohol use disorder was coded as lifetime history of Diagnostic and Statistical Manual of Mental Disorders (4th ed.; DSM–IV; American Psychiatric Association, 1994) alcohol abuse or dependence (0 = no, 1 = yes). Nicotine dependence is based upon DSM–IV criteria and reflects lifetime nicotine dependence (0 = no, 1 = yes).

Prescription drug misuse and disorder

Lifetime prescription drug misuse was coded using three items, whether the participant had ever used (a) sedatives, (b) tranquilizers, or (c) opioids (heroin is asked separately; [0 = no, 1 = yes]). Participants were instructed to respond about use that was “on your own-that is either without a doctor’s prescription; in greater amounts, more often or longer than prescribed; or for a reason other than a doctor said you should use them.” Prescription drug use disorder was coded as lifetime history of DSM-IV sedative, tranquilizer, or opioid (excluding heroin) abuse or dependence (0 = no, 1 = yes).

Data Analytic Plan

Preliminary descriptive analyses and logistic regression analyses were conducted for the current study. In order to account for the sampling design, the complex samples module for SPSS 24 was used for data analysis. First, a series of unadjusted analyses were conducted to test bivariate associations between each family variable of interest and covariates with prescription drug misuse. The same unadjusted analyses were conducted to test bivariate associations between each family variable of interest and covariates with prescription drug use disorders. Next, an overall logistic regression model was tested using all significant variables from unadjusted bivariate analyses; in these analyses, conducted separately for prescription drug misuse and prescription drug disorders, the following independent variables were modeled: biological sex, income, nativity, marital status, parental alcoholism, parental divorce before age 18, and parental death before age 18.

For lifetime prescription drug misuse, unadjusted analyses were statistically significant for all variables with the exception of parental death before age 18 (Table 2). In adjusted analyses, the overall model was statistically significant, Wald F = 578.75, p < .001 and explained between 5.7% (Cox and Snell) and 17.7% (Nagelkerke) of the variance in prescription drug misuse. Parental alcoholism (odds ratio [OR] = 1.19, 95% CI [1.07, 1.32]) and parental divorce (OR = 1.55, 95% CI [1.41, 1.71]) before the age of 18 both increased the odds of prescription drug misuse. Lifetime alcohol use disorder (OR = 3.85, 95% CI [3.46, 4.29]) and nicotine dependence (OR = 3.48, 95% CI [2.91, 4.20]) were both associated with greater odds of prescription drug misuse. In adjusted analyses, women had greater odds of prescription drug misuse (OR = 1.25, 95% CI [1.15, 1.37]). Participants born outside of the United States (OR = .66, 95% CI [.58, .74]) and those married or living as married (OR = .68, 95% CI [.60, .76]) had lower odds of prescription drug misuse. Compared with those with incomes greater than $80,000, those with lower income had reduced odds of prescription drug misuse (see Table 2 for ORs by income group).

Unadjusted and Adjusted Odds Ratios

The next series of analyses tested the associations between the family variables of interest and prescription drug use disorder. All variables were significantly related to the outcome and, thus, were retained in adjusted analyses. The overall model was statistically significant, Wald F = 1,873.24, p < .001 and explained between 3.4 (Cox and Snell) and 26.4% (Nagelkerke) of the variance in prescription drug use disorder. Parental alcoholism (OR = 1.35, 95% CI [1.07, 1.71]), parental divorce (OR = 2.43, 95% CI [1.96, 2.99]), and death of a parent (OR = 1.63, 95% CI [1.22, 2.19]) all increased the odds of a prescription drug use disorder. Lifetime alcohol use disorder (OR = 5.92, 95% CI [5.26, 6.67]) and nicotine dependence (OR = 4.57, 95% CI [3.87, 5.46] were both associated with greater odds of prescription drug misuse. Women (OR = .77, 95% CI [.64, .92]), participants born outside of the United States (OR = .52, 95% CI [.44, .62]), and those married or living as married (OR = .53, 95% CI [.41, .68]) had lower odds of a prescriptions drug use disorder. Like lifetime prescription drug misuse, those with lower incomes had lower odds of prescription drug use disorder when compared with those with incomes greater than $80, 000 (see Table 2 for odds ratios by income group).

The aim of the current study was to test the associations between family characteristics prior to age 18 and prescription drug misuse and prescription drug use disorder in a national sample of Latino adults. Prescription drug misuse is a critical public health problem and, historically, has been studied less than other substance classes such as alcohol, tobacco, and marijuana. Understanding risk and protective factors for prescription drug use is critical for the development of prevention and intervention for Latino populations. Family is a core developmental context and Familismo represents an important Latino cultural value (Santiago-Rivera et al., 2002). With respect to prescription drug misuse, it is not clear whether family variables are related in the same ways as for other substances given that attitudes regarding prescription drug use may be more permissive than for other drugs.

Results of the current study underscore the role that family disruption may play in prescription drug misuse behaviors. Parental alcoholism and parental divorce before the age of 18 both increased the odds of any prescription drug misuse as well as prescription drug use disorder even when accounting for other substance use disorders. These results are inconsistent with the adolescent literature which found no relationship between family structure and prescription drug misuse (Harrell & Broman, 2009). There are a number of reasons why this may have occurred. The NESARC is a study of adults, and thus, the prevalence of prescription drug use is greater and allowed for such associations to emerge later in participants’ lives, when access to prescription drugs may be more likely. Barrett and Turner (2006) found that intact family structure was associated with reduced risk for substance use among a sample of young adults with a large subsample of Latinos. The results of the current study are consistent with their study. With respect to parental alcoholism, the results are consistent with other research that has found broad associations between parental alcoholism and substance use disorders (Mellentin et al., 2016) and parental substance use and adolescent prescription drug misuse (Tucker et al., 2015). Death of a parent before the age of 18 increased the odds of a prescription drug use disorder. This result is consistent with previous research using a Caucasian sample of male twins that found associations between parental death and the development of alcohol use disorders (Otowa, York, Gardner, Kendler, & Hettema, 2014).

Of note, a number of control variables were also significantly related to prescription drug misuse and prescription drug use disorders in this Latino subsample. Alcohol and nicotine use disorders were robust predictors of prescription drug misuse. This finding is not surprising, given high rates of co-occurring substance use and substance use disorders (Falk, Yi, & Hiller-Sturmhöfel, 2006, 2008). Latinas were at lower odds for lifetime prescription drug disorders, but not lifetime misuse. Consistent with previous research using the NESARC that found a lower lifetime prevalence of substance use disorders among those born outside the United States (Alegria et al., 2006), participants who were born outside of the United States were also at lower odds for prescription drug misuse and disorders. One potential explanation is that those who are born outside of the United States may hold traditional values that do not promote substance use.

Participants with lower incomes also had lower odds of prescription drug misuse and disorder. More specifically, compared with those participants who reported family incomes greater than $80,000, those in every other income category had lower odds of prescription drug misuse and prescription drug misuse disorders. These results are contrary to other NESARC studies that include all racial and ethnic groups (e.g., Martins, Keyes, Storr, Zhu, & Chilcoat, 2009; Wu, Woody, Yang, & Blazer, 2010). This finding may be connected to the focus on Latino participants in the study. Lower incomes may be a proxy indicator for less acculturation to dominant United States culture which has been associated with lower rates of substance use among Latinos (Zemore, 2007). Further, while not the case for other racial and ethnic groups and prescription drug access, lower income for Latino groups may be associated with poorer access to both health care and the financial means to obtain such substances. This represents an important area of further research. Finally, consistent with previous research, participants who are married or living as married had lower odds of prescription drug misuse and disorder than those with other marital statuses. It may be that being married or in a marriagelike relationship confers a greater level of social support and connectedness to reduce risk for prescription drug misuse.

The current study should be understood in the context of its strengths and limitations. First, the NESARC is a national representative sample allowing for greater generalizability of findings to the larger Latino population in the United States. This is important, given that many studies of Latinos are from geographically restricted areas. Second, there has been a growing interest in prescription drug misuse and research is needed to identify important risk and protective factors among understudied populations such as the Latino population. The aim of the current study was to focus in on family factors as an important developmental and cultural context for this population and to tests these variables with respect to prescription drug misuse and disorder in a population of adults.

There are limitations of the current study also worth noting. The results are correlational in nature and do not allow for causal conclusions. Family factors account for a small amount of the variance in prescription drug misuse. That said, this is not atypical in risk and protective factor research for substance use and use disorders. The factors that contribute to prescription drug use come from multiple and often related contexts. The aim of the current study was to test a group of variables (family characteristics) that are culturally relevant for Latinos to inform prevention and intervention. In addition, the current study was not able to test other cultural variables such as acculturation and enculturation as the NESARC does not have such variables. Due to low prevalence, particularly for the prescription drug use disorder, the results of the current study did not allow for analyses for separate drug classes used in the composite variables.

Future research in this area should look at childhood family factors using longitudinal methods as well as test important mediators and moderators of these relationships. In addition, the roles that culture and cultural values play in substance use behaviors is relevant to Latinos living in the United States. Current models of acculturation are tuned into behavioral and values shifts as well as the receiving environment for Latino populations born outside of the United States (Schwartz, Unger, Zamboanga, & Szapocznik, 2010). These cultural contexts are influential in drug use attitudes and behaviors and are an important area of further research with respect to prescription drug misuse. For example, future research might test how family structure (including parental divorce) might moderate the relationship between the Latino cultural value of Familismo and prescription drug misuse. In sum, the relationships between family structure, family related values and substance use behaviors among Latinos are complex and warrant further research to inform the development of prevention and intervention strategies.

Alegria, M., Canino, G., Stinson, F. S., & Grant, B. F. (2006). Nativity and DSM–IV psychiatric disorders among Puerto Ricans, Cuban Americans, and non-Latino whites in the United States: Results from the national epidemiologic survey on alcohol and related conditions. The Journal of Clinical Psychiatry, 67, 56–65. 10.4088/JCP.v67n0109

American Psychiatric Association. (1994). Diagnostic and statistical manual of mental disorders (4th ed.). Washington, DC: Author.

Barrett, A. E., & Turner, R. J. (2006). Family structure and substance use problems in adolescence and early adulthood: Examining explanations for the relationship. Addiction, 101, 109–120. 10.1111/j.1360-0443.2005.01296.x

Cai, R., Crane, E., Poneleit, K., & Paulozzi, L. (2010). Emergency department visits involving nonmedical use of selected prescription drugs—United States, 2004–2008. Morbidity and Mortality Weekly Report (Vol. 59, pp. 705–709). Atlanta, GA: Centers for Disease Control and Prevention. Retrieved from http://www.cdc.gov.offcampus.lib.washington.edu/mmwr/preview/mmwrhtml/mm5923a1.htm

Centers for Disease Control and Prevention. (2016). Prescription opioid overdose data. Retrieved from www.cdc.gov/drugoverdose/data/overdose.html

Chen, C. M., Yi, H., Falk, D. E., Stinson, F. S., Dawson, D. A., & Grant, B. F. (2006). Alcohol use and alcohol use disorders in the United States: Main findings from the 2001–2002 National Epidemiologic Survey on Alcohol and Related Conditions (NESARC), (NIH Publication No. 05–5737). Bethesda, MD: National Institute on Alcohol Abuse and Alcoholism.

Coatsworth, J. D., Pantin, H., & Szapocznik, J. (2002). Familias Unidas: A family-centered ecodevelopmental intervention to reduce risk for problem behavior among Hispanic adolescents. Clinical Child and Family Psychology Review, 5, 113–132. 10.1023/A:1015420503275

Conn, B. M., & Marks, A. K. (2014). Ethnic/racial differences in peer and parent influence on adolescent prescription drug misuse. Journal of Developmental and Behavioral Pediatrics, 35, 257–265. 10.1097/DBP.0000000000000058

Falk, D. E., Yi, H. Y., & Hiller-Sturmhöfel, S. (2006). An epidemiologic analysis of co-occurring alcohol and tobacco use and disorders: Findings from the National Epidemiologic Survey on Alcohol and Related Conditions. Alcohol Research & Health, 29, 162–171.

Falk, D., Yi, H. Y., & Hiller-Sturmhöfel, S. (2008). An epidemiologic analysis of co-occurring alcohol and drug use and disorders: Findings from the National Epidemiologic Survey on Alcohol and Related Conditions (NESARC). Alcohol Research & Health, 31, 100–110.

Ford, J. A., & Rigg, K. K. (2015). Racial/ethnic differences in factors that place adolescents at risk for prescription opioid misuse. Prevention Science, 16, 633–641. 10.1007/s11121-014-0514-y

Grant, B. F., & Dawson, D. A. (2006). Introduction to the National Epidemiologic Survey on Alcohol and Related Conditions. Alcohol Research & Health, 29, 74–78.

Harrell, Z. A., & Broman, C. L. (2009). Racial/ethnic differences in correlates of prescription drug misuse among young adults. Drug and Alcohol Dependence, 104, 268–271. 10.1016/j.drugalcdep.2009.05.017

Johnston, L. D., O’Malley, P. M., Miech, R. A., Bachman, J. G., & Schulenberg, J. E. (2016). Demographic subgroup trends among adolescents in the use of various licit and illicit drugs, 1975–2015 (Monitoring the Future Occasional Paper No. 86). Ann Arbor, MI: Institute for Social Research, University of Michigan.

Knight, G. P., Gonzales, N. A., Saenz, D. S., Bonds, D. D., Germán, M., Deardorff, J., . . .Updegraff, K. A. (2010). The Mexican American Cultural Values Scale for Adolescents and Adults. The Journal of Early Adolescence, 30, 444–481. 10.1177/0272431609338178

Martins, S. S., Keyes, K. M., Storr, C. L., Zhu, H., & Chilcoat, H. D. (2009). Pathways between nonmedical opioid use/dependence and psychiatric disorders: Results from the National Epidemiologic Survey on Alcohol and Related Conditions. Drug and Alcohol Dependence, 103, 16–24. 10.1016/j.drugalcdep.2009.01.019

Mellentin, A. I., Brink, M., Andersen, L., Erlangsen, A., Stenager, E., Bjerregaard, L. B., & Christiansen, E. (2016). The risk of offspring developing substance use disorders when exposed to one versus two parent(s) with alcohol use disorder: A nationwide, register-based cohort study. Journal of Psychiatric Research, 80, 52–58. 10.1016/j.jpsychires.2016.06.001

Moore, A. A., Karno, M. P., Grella, C. E., Lin, J. C., Warda, U., Liao, D. H., & Hu, P. (2009). Alcohol, tobacco, and nonmedical drug use in older U.S. adults: Data from the 2001/02 National Epidemiologic Survey on Alcohol and Related Conditions. Journal of the American Geriatrics Society, 57, 2275–2281. 10.1111/j.1532-5415.2009.02554.x

Otowa, T., York, T. P., Gardner, C. O., Kendler, K. S., & Hettema, J. M. (2014). The impact of childhood parental loss on risk for mood, anxiety and substance use disorders in a population-based sample of male twins. Psychiatry Research, 220, 404–409. 10.1016/j.psychres.2014.07.053

Santiago-Rivera, A. L., Arredondo, P., & Gallardo-Cooper, M. (2002). Counseling Latinos and la familia. Thousand Oaks, CA: Sage.

Schwartz, S. J., Unger, J. B., Zamboanga, B. L., & Szapocznik, J. (2010). Rethinking the concept of acculturation: Implications for theory and research. American Psychologist, 65, 237–251. 10.1037/a0019330

Sher, K. J. (1991). Children of alcoholics: A critical appraisal of theory and research. Chicago, IL: University of Chicago Press.

Stepler, R., & Brown, A. (2016). Statistical portrait of Hispanics in the United States. Retrieved from http://www.pewhispanic.org/2016/04/19/statistical-portrait-of-hispanics-in-the-united-states-key-charts/

Substance Abuse and Mental Health Services Administration (SAMHSA). (2015). Behavioral health trends in the United States: Results from the 2014 National Survey on Drug Use and Health. Rockville, MD: Author. Retrieved from www.samhsa.gov/data/sites/default/files/NSDUH-FRR1-2014/NSDUH-FRR1-2014.pdf

Substance Abuse and Mental Health Services Administration, Center for Behavioral Health Statistics and Quality. (2013). The NSDUH report: Nonmedical use of prescription-type drugs, by county type. Rockville, MD: Author. Retrieved from www.samhsa.gov: http://www.samhsa.gov/data/2k13/NSDUH098/sr098-UrbanRuralRxMisuse.htm

Tucker, J. S., Ewing, B. A., Miles, J. N. V., Shih, R. A., Pedersen, E. R., & D’Amico, E. J. (2015). Predictors and consequences of prescription drug misuse during middle school. Drug and Alcohol Dependence, 156, 254–260. 10.1016/j.drugalcdep.2015.09.018

Volkow, N. D. (2006). Hispanic drug abuse research: Challenges and opportunities. Drug and Alcohol Dependence, 84, S4–S7. 10.1016/j.drugalcdep.2006.05.002

Waldron, M., Grant, J. D., Bucholz, K. K., Lynskey, M. T., Slutske, W. S., Glowinski, A. L., . . .Heath, A. C. (2014). Parental separation and early substance involvement: Results from children of alcoholic and cannabis dependent twins. Drug and Alcohol Dependence, 134, 78–84. 10.1016/j.drugalcdep.2013.09.010

Waldron, M., Vaughan, E. L., Bucholz, K. K., Lynskey, M. T., Sartor, C. E., Duncan, A. E., . . .Heath, A. C. (2014). Risks for early substance involvement associated with parental alcoholism and parental separation in an adolescent female cohort. Drug and Alcohol Dependence, 138, 130–136. 10.1016/j.drugalcdep.2014.02.020

Windle, M., & Searles, J. S. (1990). Children of alcoholics: Critical perspectives. New York, NY: Guilford Press.

Wu, L. T., Woody, G. E., Yang, C., & Blazer, D. G. (2010). Subtypes of nonmedical opioid users: Results from the national epidemiologic survey on alcohol and related conditions. Drug and Alcohol Dependence, 112, 69–80. 10.1016/j.drugalcdep.2010.05.013

Zemore, S. E. (2007). Acculturation and alcohol among Latino adults in the United States: A comprehensive review. Alcoholism: Clinical and Experimental Research, 31, 1968–1990. 10.1111/j.1530-0277.2007.00532.x

Submitted: October 19, 2016 Revised: March 6, 2017 Accepted: March 19, 2017

Coping-Motivated Marijuana Use Correlates With DSM-5 Cannabis Use Disorder and Psychological Distress Among Emerging Adults

Acknowledgement: This study was funded by NIAAA Grant R01 AA020509. Trial registered at clinicaltrials.gov as NCT01473719.

Evolving social attitudes toward marijuana (MJ) have led to legalization of its use for medical and recreational purposes in some U.S. states. Over the past 30 years, disapproval of MJ use has decreased across birth cohorts (Keyes et al., 2011), with emerging adults (ages 18–25 years) being most accepting of use. Compared to other age cohorts, emerging adults (Arnett, 2001) also have the highest rates of MJ use, and substance use disorders peak during this period (Substance Abuse and Mental Health Services Administration [SAMHSA], 2013). Indeed, studies have shown that nearly 10% of college-based emerging adults meet criteria for an MJ use disorder (Caldeira, Arria, O’Grady, Vincent, & Wish, 2008; Caldeira et al., 2009).

Reasons for substance use can vary between and within individuals (Cooper, 1994). Common motives are to cope with negative emotions or distress (e.g., “to forget my worries”), to conform (e.g., “because I felt pressure from others who do it”), for enhancement purposes (e.g., “because I like the feeling”), for expansion (e.g., “to expand my awareness”), and for social purposes (e.g., “it’s what I do with friends”). Conformity-motivated use is driven by a desire to reduce social exclusion, enhancement-motivated use is described as being driven by a desire for excitement or joy, expansion-motivated use relates to seeking cognitive or perceptual enhancement of experiences, and social-motivated use seeks to facilitate social cohesion (Simons, Correia, Carey, & Borsari, 1998; Simons, Gaher, Correia, Hansen, & Christopher, 2005). According to the stress-coping model (Wills & Shiffman, 1985), people may also consume substances as a coping response to stress, with the substance being used to engender positive affect and/or decrease an aversive mood.

Enhancement, expansion, and social motives are positively associated with MJ use but less related to negative outcomes such as MJ-related problems or psychological distress (Bonn-Miller, Zvolensky, & Bernstein, 2007; Brodbeck, Matter, Page, & Moggi, 2007). However, individuals with MJ use disorders have higher rates of enhancement-motivated MJ use compared to more casual MJ users (Bonn-Miller & Zvolensky, 2009). Although using MJ to conform is associated with social anxiety symptoms (Buckner, Bonn-Miller, Zvolensky, & Schmidt, 2007), it has been found to negatively correlate with recent MJ use (Bonn-Miller, Zvolensky, & Bernstein, 2007).

Although most emerging adults report using MJ primarily for enhancement or social reasons (Lee, Neighbors, & Woods, 2007), those who endorse greater MJ use to cope with distress may represent a subgroup trying to manage more severe mental health problems and, in doing so, may be at risk for MJ-related problems. Emerging adults may be more likely than other adults to use MJ to cope with psychological distress (Buckner, 2013). Using MJ to cope with distress is associated with negative affect, anxious arousal, and depressive symptoms (Beck et al., 2009; Mitchell, Zvolensky, Marshall, Bonn-Miller, & Vujanovic, 2007). Among emerging adults with a history of trauma, coping-motivated use, but not other motives, is associated with posttraumatic stress symptoms (Bonn-Miller, Vujanovic, Feldner, Bernstein, & Zvolensky, 2007). Using MJ to cope with negative emotions is also uniquely associated with emotional dysregulation (Bonn-Miller, Vujanovic, & Zvolensky, 2008) and social anxiety symptoms (Buckner et al., 2007) relative to other motives. Although these studies indicate that emerging adults who use MJ to cope experience psychological distress, they are limited by the exclusion of individuals with current Axis I psychopathology (Bonn-Miller, Vujanovic, et al., 2007) and restriction to college students (Buckner et al., 2007). A more representative sample of emerging adults who use MJ to cope with distress is needed to better understand the relationship among these factors.

Coping-motivated MJ use in emerging adults is also associated with more persistent use (Patrick, Schulenberg, O’Malley, Johnston, & Bachman, 2011; Patrick, Schulenberg, O’Malley, Maggs, et al., 2011; Titus, Godley, & White, 2007). Persistent use can lead to MJ-related problems, particularly for individuals who start using earlier in life (Anthony & Petronis, 1995). Among emerging adult MJ users, those who use to cope with distress are at increased risk for MJ-related problems (Buckner, 2013; Lee et al., 2007) and are more likely to meet criteria from the fourth edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV; American Psychiatric Association, 2000) for MJ dependence (Bonn-Miller & Zvolensky, 2009). Yet the DSM-IV had inadequate clinical utility in discriminating MJ problem severity among emerging adults (Martin, Chung, Kirisci, & Langenbucher, 2006). To our knowledge, no study has investigated the link between using MJ to cope with distress and cannabis use disorder as defined in the fifth edition of the DSM (DSM-5; American Psychiatric Association, 2013).

In this study, we examined the association of motivations for using MJ (social, enhancement, and coping) in emerging adults with four measures of MJ-related problem severity and psychological distress: (a) meeting DSM-5 criteria for cannabis use disorder, (b) MJ-related problem severity, (c) depressive symptomatology, and (d) perceived stress. We hypothesized that coping-motivated use would be more strongly associated with these adverse outcomes than social- or enhancement-motivated use.

Participants

Participants were recruited for a large study on health behaviors among emerging adults who use MJ or alcohol through advertisements online, in local college newspapers, on public transportation, and on commercial radio in Rhode Island. After a telephone screen, eligible individuals were invited for a compensated ($40) in-person interview and free sexually transmitted infection testing. The study was approved by the Butler Hospital Institutional Review Board.

Eligibility criteria included being 18–25 years old, drinking alcohol and/or using MJ in the past month, being sexually active in the past 6 months, not having suicidal ideation in the past 2 weeks, and living within 30 minutes of the research site. Of the 1,621 individuals screened by phone, 689 were ineligible. The remaining 932 eligible persons were invited for an interview, and 533 were either not interested or did not keep a scheduled baseline appointment. In total, 399 individuals completed baseline interviews, after which 17 persons were found to be ineligible. For the present analysis, we included data only from individuals who reported using MJ in the past 30 days (n = 288).

Measures

Frequency of cigarette smoking

Frequency of cigarette smoking was assessed using an item from the Fagerström Test for Nicotine Dependence (Heatherton, Kozlowski, Frecker, & Fagerström, 1991): “How many cigarettes do you smoke per day?”

Marijuana Problem Scale

The Marijuana Problem Scale (MPS; Stephens, Roffman, & Curtin, 2000; Stephens et al., 2004) is a reliable and valid measure (score range 0–48) of 19 problems directly related to MJ use, ranging from losing a job to having withdrawal symptoms to having problems in one’s family.

Patient Health Questionnaire–9

The Patient Health Questionnaire–9 (PHQ-9; Kroenke, Spitzer, & Williams, 2001) is a validated and reliable nine-item measure (score range 0–27) of depressive symptoms.

Perceived Stress Scale–4

The Perceived Stress Scale–4 (PSS-4; Cohen, Kamarck, & Mermelstein, 1983) is a four-item measure (score range 0–16) that assesses the degree to which individuals perceive their environment and experiences as stressful.

Reasons for Marijuana Use

We adapted the Reasons for Drinking measure developed by Cooper, Russell, Skinner, and Windle (1992) for this study to examine MJ use motives. This measure has three subscales: (a) coping (sample item, “Because it helped when you felt depressed or nervous”), (b) enhancement (“Because it’s exciting”), and (c) social (“To be sociable”). The Reasons for Marijuana Use subscales had a possible range of 1–4, corresponding to never/almost never, sometimes, often, and almost always. In this sample, internal consistency reliabilities were .84, .84, and .73 for the coping, enhancement, and social scales, respectively. Product-moment correlations between the subscales ranged from .45–.47.

Structured Interview for the DSM-IV—cannabis abuse and dependence modules

The Structured Interview for the DSM-IV (First, Spitzer, Gibbon, & Williams, 1996) is the most widely used, reliable, and well-validated, structured clinical assessment tool for DSM-IV diagnostic criteria. To assess the craving criterion under DSM-5 defined cannabis use disorder, we asked all participants, “In the past 3 months, have you often had cravings or strong desires or urges to use marijuana?” Participants endorsing two or more of the abuse or dependence items, including our additional craving question, met criteria for DSM-5 cannabis use disorder. Severity of cannabis use disorder was coded by number of criteria endorsed: no disorder = zero or one symptom, mild = two or three symptoms, moderate = four or five symptoms, and severe = six or more symptoms (Hasin et al., 2013).

Timeline Follow-Back Measure

The Timeline Follow-Back Measure (TLFB; Sobell & Sobell, 1996) is a semistructured interview that uses a calendar-guided approach (Fals-Stewart, O’Farrell, Freitas, McFarlin, & Rutigliano, 2000) and assesses alcohol and MJ use in the past 30 days.

Data Analysis

Descriptive statistics summarize the characteristics of the sample. Our primary focus is on the associations of using MJ for coping, socialization, and enhancement with indicators of MJ use severity and problems. We also examined associations with measures of psychological well-being. Background characteristics included as covariates were age, gender, ethnoracial group, employment status, education, alcohol use frequency (past 30 days), MJ use frequency (past 30 days), and number of cigarettes smoked per day (past 30 days). Associations were estimated in a seemingly unrelated regression framework (Zellner, 1962) using Mplus 5.1 (Muthén & Muthén, 2008). This method assumes error terms are correlated across equations and parameter estimates are more efficient than equation-by-equation estimation. The interpretation of estimated coefficients is identical to single-equation regression models. Prior to analysis, all continuous variables were standardized to zero-mean and unit variance. For the equations with continuous dependent variables (PHQ-9, PSS-4, and MPS), the coefficients reported for continuous factors are fully standardized regression coefficients, and the coefficients for the categorical factors are y-standardized. Associations with meeting criteria for DSM-5 cannabis use disorder are reported as odds ratios. Parameters and inferential statistics were estimated using maximum likelihood with robust standard errors (MLR in Mplus). All above-described covariates and the three motivation-to-use MJ subscales were entered simultaneously in the multivariate models. An indicator variable contrasting non-Latino Whites to all other racial or ethnic identifications was used in analyses.

Of the 288 emerging adults who reported MJ use in the past 30 days, mean age was 21.2 (SD = 2.1) years, 135 (51.7%) were male, and 187 (64.9%) were non-Latino White (see Table 1). On average, participants used alcohol and MJ on 26.7% (SD = 18.0%) and 52.5% (SD = 38.1%) of TLFB days, respectively. More than two thirds (70.5%) reported no cigarette smoking in the 30 days prior to baseline.

Background Characteristics and Descriptive Statistics (n = 288)

After adjusting for other variables in the model, including using MJ for enhancement and social reasons, using MJ to cope with distress was positively and significantly associated with meeting DSM-5 diagnostic criteria for cannabis use disorder (OR = 1.85, 95% CI [1.31, 2.62], p < .01). As a supplementary analysis, we estimated a parallel ordinal logit regression model in which cannabis use disorder severity was regressed on the reasons to use indices and all covariates described in Table 2; results were consistent with those reported for the dichotomized outcome. Using MJ to cope was associated positively and significantly with cannabis use disorder severity (OR = 1.59, 95% CI [1.15, 2.19], p < .05). Neither using for social reasons (OR = 1.46, 95% CI [0.98, 2.18], p > .05) nor using for enhancement (OR = 0.94, 95% CI [0.66, 1.34], p > .05) was associated significantly with cannabis use disorder severity. Results were the same when analyzing the unique associations of the three reasons for using MJ and cannabis use disorder based on number of criteria endorsed.

Seemingly Unrelated Regression Model Estimating the Adjusted Association of Using Marijuana to Cope, to Socialize, and for Enhancement on Various Measures (n = 288)

Using MJ to cope was also significantly associated with MJ problem severity (b = .41, 95% CI [.24, .57], p < .01), depressive symptomatology (b = .36, 95% CI [.23, .49], p < .01), and perceived stress (b = .37, 95% CI [.22, .51], p < .01; see Table 2). These multivariate models estimated the effects of the other reasons for using MJ subscales, revealing that using MJ to socialize or for enhancement purposes was not uniquely associated significantly with any of the dependent variables (see Table 2).

This study found that among emerging adults who use MJ, use to cope with distress is positively and significantly associated with having a DSM-5 cannabis use disorder. Using MJ for enhancement or social purposes did not uniquely account for a significant proportion of variance in this outcome. Moreover, coping-motivated use, but not social- or enhancement-motivated use, is associated with MJ-related problems in this group. Using MJ to cope with negative emotions among emerging adults also appears to be uniquely associated with psychiatric symptoms, as measured by severity of depressive symptoms and degree of perceived stress, consistent with prior research (Mitchell et al., 2007). These data are the first to demonstrate the confluence of cannabis use disorder, MJ-related problems, and psychiatric symptoms in the same sample. In addition, a significant shortcoming of previous work was the exclusion of individuals who met DSM diagnostic criteria for Axis I psychopathology (Bonn-Miller, Vujanovic, et al., 2007; Bonn-Miller et al., 2008; Bonn-Miller & Zvolensky, 2009), a meaningful omission given the concern for mental health issues in these individuals.

The acceptability of MJ use is growing in emerging adults (Keyes et al., 2011), a high-risk group for substance use disorders (SAMHSA, 2013). Although prior studies have shown an association between coping-motivated MJ use and a variety of negative psychological factors (e.g., Mitchell et al., 2007), little research has compared the relationship of coping-motivated use, relative to social- and enhancement-based use, to MJ-related problems and psychological variables. These are also the first results linking using MJ to cope with psychological distress to the newly defined DSM-5 cannabis use disorder. This new diagnostic category represents an important streamlining of the DSM’s cannabis abuse and dependence diagnoses while incorporating a severity dimension. This new approach is particularly relevant to clinicians working with emerging adults, because the DSM-IV classification system poorly quantified severity of use in this age group (Martin et al., 2006).

This study had limitations. First, our primary measure of MJ use motives was adapted from an alcohol scale. Moreover, we did not measure conformity- or expansion-motivated use; these would be important to include in future research. Second, although coping-motivated use was significantly associated with negative psychological variables, our estimated standardized effect sizes suggest that coping-motivated use might not be the only factor associated with these outcomes. Third, the sample was limited to emerging adults who had past-month MJ use, were not seeking treatment, and were sexually active. Although the sample had the strengths of not excluding those with Axis I psychopathology, being ethnically/racially diverse, being nearly half female, and including a substantial number of emerging adults not currently in school (41%), it was not an epidemiological sample. Furthermore, > 50% of potential participants declined to participate in the study. Fourth, we did not use a diagnostic measure to assess the presence of major depressive disorder. Fifth, MJ-induced anxiety is one of the most commonly reported acute symptoms of MJ use (Crippa et al., 2009). Thus, it is possible that self-reported use of MJ to cope with distress is confounded by MJ-triggered symptoms. Finally, given the cross-sectional nature of the data, we were unable to examine the temporal relationship between coping-motivated use and psychological distress. Still, our findings suggest that assessment of use of MJ to manage psychological distress may be clinically important and, if found, signal the importance of a broad and careful mental health assessment.

These results raise the question about how coping-motivated MJ use might improve or worsen one’s well-being. Although MJ may be perceived as beneficial in ameliorating symptoms of emotional distress, long-term MJ use for these purposes has been associated with deleterious consequences (Patrick, Schulenberg, O’Malley, Johnston, et al., 2011). More longitudinal research will be needed to examine if, despite being used with the intention of mitigating distress, coping-motivated use may actually worsen psychological health.

MJ use is becoming more socially acceptable and common in emerging adults. These results help clinicians identify MJ-using individuals who are likely to also have psychological distress symptoms. It appears that there is an important subset of emerging adults who use MJ for coping purposes, and these individuals are at risk for a variety of MJ-related problems. Clinicians working with patients who use MJ to cope with negative emotions face the challenge of confronting misconceptions about the perceived benefit of using MJ to “treat” distress. If MJ users continue to be reluctant to engage in drug counseling or to reduce use, despite having substance-related problems, clinicians must become more open to providing treatment such as alternative coping strategies for what these users might be more motivated to change, namely, their symptoms of psychological distress.

Anthony, J. C., & Petronis, K. R. (1995). Early-onset drug use and risk of later drug problems. Drug and Alcohol Dependence, 40, 9–15. 10.1016/0376-8716(95)01194-3

American Psychiatric Association. (2000). Diagnostic and statistical manual of mental disorders (4th ed.). Washington, DC: Author.

American Psychiatric Association. (2013). Diagnostic and statistical manual of mental disorders (5th ed.). Arlington, VA: American Psychiatric Publishing.

Arnett, J. J. (2001). Conceptions of the transition to adulthood: Perspectives from adolescence through midlife. Journal of Adult Development, 8, 133–143. 10.1023/A:1026450103225

Beck, K. H., Caldeira, K. M., Vincent, K. B., O’Grady, K. E., Wish, E. D., & Arria, A. M. (2009). The social context of cannabis use: Relationship to cannabis use disorders and depressive symptoms among college students. Addictive Behaviors, 34, 764–768. 10.1016/j.addbeh.2009.05.001

Bonn-Miller, M. O., Vujanovic, A. A., Feldner, M. T., Bernstein, A., & Zvolensky, M. J. (2007). Posttraumatic stress symptom severity predicts marijuana use coping motives among traumatic event-exposed marijuana users. Journal of Traumatic Stress, 20, 577–586. 10.1002/jts.20243

Bonn-Miller, M. O., Vujanovic, A. A., & Zvolensky, M. J. (2008). Emotional dysregulation: Association with coping-oriented marijuana use motives among current marijuana users. Substance Use & Misuse, 43, 1653–1665. 10.1080/10826080802241292

Bonn-Miller, M. O., & Zvolensky, M. J. (2009). An evaluation of the nature of marijuana use and its motives among young adult active users. The American Journal on Addictions, 18, 409–416. 10.3109/10550490903077705

Bonn-Miller, M. O., Zvolensky, M. J., & Bernstein, A. (2007). Marijuana use motives: Concurrent relations to frequency of past 30-day use and anxiety sensitivity among young adult marijuana smokers. Addictive Behaviors, 32, 49–62. 10.1016/j.addbeh.2006.03.018

Brodbeck, J., Matter, M., Page, J., & Moggi, F. (2007). Motives for cannabis use as a moderator variable of distress among young adults. Addictive Behaviors, 32, 1537–1545. 10.1016/j.addbeh.2006.11.012

Buckner, J. D. (2013). College cannabis use: The unique roles of social norms, motives, and expectancies. Journal of Studies on Alcohol and Drugs, 74, 720–726.

Buckner, J. D., Bonn-Miller, M. O., Zvolensky, M. J., & Schmidt, N. B. (2007). Marijuana use motives and social anxiety among marijuana-using young adults. Addictive Behaviors, 32, 2238–2252. 10.1016/j.addbeh.2007.04.004

Caldeira, K. M., Arria, A. M., O’Grady, K. E., Vincent, K. B., & Wish, E. D. (2008). The occurrence of cannabis use disorders and other cannabis-related problems among first-year college students. Addictive Behaviors, 33, 397–411. 10.1016/j.addbeh.2007.10.001

Caldeira, K. M., Kasperski, S. J., Sharma, E., Vincent, K. B., O’Grady, K. E., Wish, E. D., & Arria, A. M. (2009). College students rarely seek help despite serious substance use problems. Journal of Substance Abuse Treatment, 37, 368–378. 10.1016/j.jsat.2009.04.005

Cohen, S., Kamarck, T., & Mermelstein, R. (1983). A global measure of perceived stress. Journal of Health and Social Behavior, 24, 385–396. 10.2307/2136404

Cooper, M. L. (1994). Motivations for alcohol use among adolescents: Development and validation of a four-factor model. Psychological Assessment, 6, 117–128. 10.1037/1040-3590.6.2.117

Cooper, M. L., Russell, M., Skinner, J., & Windle, M. (1992). Development and validation of a three-dimensional measure of drinking motives. Psychological Assessment, 4, 123–132. 10.1037/1040-3590.4.2.123

Crippa, J. A., Zuardi, A. W., Martín-Santos, R., Bhattacharyya, S., Atakan, Z., McGuire, P., & Fusar-Poli, P. (2009). Cannabis and anxiety: A critical review of the evidence. Human Psychopharmacology: Clinical and Experimental, 24, 515–523. 10.1002/hup.1048

Fals-Stewart, W., O’Farrell, T. J., Freitas, T. T., McFarlin, S. K., & Rutigliano, P. (2000). The timeline followback reports of psychoactive substance use by drug-abusing patients: Psychometric properties. Journal of Consulting and Clinical Psychology, 68, 134–144.

First, M. B., Spitzer, R. L., Gibbon, M., & Williams, J. B. W. (1996). Structured Clinical Interview for DSM-IV Axis I Disorders (SCID-IV). Washington, DC: American Psychiatric Press.

Hasin, D. S., O’Brien, C. P., Auriacombe, M., Borges, G., Bucholz, K., Budney, A., . . .Grant, B. F. (2013). DSM-5 criteria for substance use disorders: Recommendations and rationale. The American Journal of Psychiatry, 170, 834–851. 10.1176/appi.ajp.2013.12060782

Heatherton, T. F., Kozlowski, L. T., Frecker, R. C., & Fagerström, K. O. (1991). The Fagerström Test for Nicotine Dependence: A revision of the Fagerström Tolerance Questionnaire. British Journal of Addiction, 86, 1119–1127. 10.1111/j.1360-0443.1991.tb01879.x

Keyes, K. M., Schulenberg, J. E., O’Malley, P. M., Johnston, L. D., Bachman, J. G., Li, G., & Hasin, D. (2011). The social norms of birth cohorts and adolescent marijuana use in the United States, 1976–2007. Addiction, 106, 1790–1800. 10.1111/j.1360-0443.2011.03485.x

Kroenke, K., Spitzer, R. L., & Williams, J. B. (2001). The PHQ-9: Validity of a brief depression severity measure. Journal of General Internal Medicine, 16, 606–613. 10.1046/j.1525-1497.2001.016009606.x

Lee, C. M., Neighbors, C., & Woods, B. A. (2007). Marijuana motives: Young adults’ reasons for using marijuana. Addictive Behaviors, 32, 1384–1394. 10.1016/j.addbeh.2006.09.010

Martin, C. S., Chung, T., Kirisci, L., & Langenbucher, J. W. (2006). Item response theory analysis of diagnostic criteria for alcohol and cannabis use disorders in adolescents: Implications for DSM-V. Journal of Abnormal Psychology, 115, 807–814. 10.1037/0021-843X.115.4.807

Mitchell, H., Zvolensky, M. J., Marshall, E. C., Bonn-Miller, M. O., & Vujanovic, A. A. (2007). Incremental validity of coping-oriented marijuana use motives in the prediction of affect-based psychological vulnerability. Journal of Psychopathology and Behavioral Assessment, 29, 277–288. 10.1007/s10862-007-9047-z

Muthén, L. K., & Muthén, B. O. (2008). Mplus user’s guide (5th ed.). Los Angeles, CA: Muthen & Muthen.

Patrick, M. E., Schulenberg, J. E., O’Malley, P. M., Johnston, L. D., & Bachman, J. G. (2011). Adolescents’ reported reasons for alcohol and marijuana use as predictors of substance use and problems in adulthood. Journal of Studies on Alcohol and Drugs, 72, 106–116.

Patrick, M. E., Schulenberg, J. E., O’Malley, P. M., Maggs, J. L., Kloska, D. D., Johnston, L. D., & Bachman, J. G. (2011). Age-related changes in reasons for using alcohol and marijuana from ages 18 to 30 in a national sample. Psychology of Addictive Behaviors, 25, 330–339. 10.1037/a0022445

Substance Abuse and Mental Health Services Administration (SAMHSA). (2013). Results from the 2012 National Survey on Drug Use and Health: Summary of National Findings. Rockville, MD: Author. Retrieved from [http://www.samhsa.gov/data/NSDUH/2012SummNatFindDetTables/]

Simons, J. S., Correia, C. J., Carey, K. B., & Borsari, B. E. (1998). Validating a five-factor marijuana motives measure: Relations with use, problems, and alcohol motives. Journal of Counseling Psychology, 45, 265–273. 10.1037/0022-0167.45.3.265

Simons, J. S., Gaher, R. M., Correia, C. J., Hansen, C. L., & Christopher, M. S. (2005). An affective-motivational model of marijuana and alcohol problems among college students. Psychology of Addictive Behaviors, 19, 326–334. 10.1037/0893-164X.19.3.326

Sobell, L. C., & Sobell, M. B. (1996). Timeline Followback user’s guide: A calendar method for assessing alcohol and drug use. Toronto, Ontario, Canada: Addiction Research Foundation.

Stephens, R. S., Roffman, R. A., & Curtin, L. (2000). Comparison of extended versus brief treatments for marijuana use. Journal of Consulting and Clinical Psychology, 68, 898–908. 10.1037/0022-006X.68.5.898

Stephens, R. S., Roffman, R. A., Fearer, S. A., Williams, C., Picciano, J. F., & Burke, R. S. (2004). The Marijuana Check-up: Reaching users who are ambivalent about change. Addiction, 99, 1323–1332. 10.1111/j.1360-0443.2004.00832.x

Titus, J. C., Godley, S. H., & White, M. K. (2007). A post-treatment examination of adolescents’ reasons for starting, quitting, and continuing the use of drugs and alcohol. Journal of Child & Adolescent Substance Abuse, 16, 31–49. 10.1300/J029v16n02_02

Wills, T. A., & Shiffman, S. (1985). Coping and substance use: A conceptual framework. In T. A.Wills & S.Shiffman (Eds.), Coping and substance use (pp. 3–24). Orlando, FL: Academic Publishing.

Zellner, A. (1962). An efficient method of estimating seemingly unrelated regression equations and tests for aggregation bias. Journal of the American Statistical Association, 57, 348–368. 10.1080/01621459.1962.10480664

Submitted: December 8, 2014 Revised: March 2, 2015 Accepted: March 9, 2015

Correlates of Recent Drug Use Among Victimized Women on Probation and Parole

Acknowledgement: The research described here was supported, in part, by a grant from the National Institute on Drug Abuse (R01DA027981). Special thanks to all the women who have participated in this research. Additional gratitude is expressed to Robin Cook, Amy Brooks, and the Kentucky Department of Corrections, Division of Probation and Parole, for their assistance.

Drug use is a serious threat to public health and functioning among women in the criminal justice system. In fact, drug use and its associated legal penalties have fueled the growth of women in the justice system over the past 30 years. In 1980, as the War on Drugs accelerated, 13,258 women were incarcerated in state or federal prison. By 2011, that number had risen to 111,387, representing a staggering 740% increase in the number of women in prison in the United States (Carson & Sabol, 2012; The Sentencing Project, 2012). Although attention most frequently focuses on individuals who are incarcerated, the majority of justice-involved women are supervised in the community (e.g., probation, parole). Thus, as the number of women who are incarcerated has grown, so too has the number of women involved in community-based sanctions. In 1990, approximately 481,000 women were on probation in the United States. By 2011, that number had increased to nearly 1 million (Glaze, 2002; Department of Justice, 2001; Maruschak & Parks, 2012). The most recent data available indicate that 1 in every 89 women in the U.S. is under some form of correctional authority, with approximately 85% assigned to probation or parole (Pew Center on the States, 2009).

Drug-related offenses and/or drug-involvement are common among the majority of justice-involved women. Female jail and prison inmates are more likely than their male counterparts to be incarcerated for a drug offense (Carson & Sabol, 2012; James, 2004). Among women, approximately 29% of jail inmates and 25% of state prisoners have been charged with drug offenses, compared with 24% and 17%, respectively, for men in jail and state prisons (Carson & Sabol, 2012; James, 2004). In addition to the high prevalence of drug-related charges, many women in the justice system also use drugs. Data from the Arrestee Drug Abuse Monitoring Program (ADAM) found that among female arrestees, 63% tested positive for use of at least one of the following substances: cocaine, marijuana, opiates, methamphetamine or PCP (National Institute of Justice, 2003). Additionally, in a recent study of female jail detainees from across the U.S., 83% experienced a substance use disorder at some point in their lives and 53% met the diagnostic criteria for a substance use disorder in the past year (Lynch, DeHart, Belknap, & Green, 2012). In comparison, only 5.9% of females over age 12 in the general population report any substance use or dependence in the past year (5.0% for alcohol and 2.0% for other drugs; Substance Abuse and Mental Health Services Administration, 2010).

Research across numerous populations has found a consistent association between women’s substance use and psychological distress (e.g., depression, anxiety, PTSD; Chilcoat & Breslau, 1998a; Chilcoat & Breslau, 1998b; Conway, Compton, Stinson, & Grant, 2006; Cottler, Compton, Mager, Spitznagel, & Janca, 1992; Dansky et al., 1995; DeHart, Lynch, Belknap, Dass-Brailsford, & Green, 2013; Hien et al., 2010; Jané-Llopis & Matytsina, 2006; Kessler, Chiu, Demler, Merikangas, & Walters, 2005; Schiff, El-Bassel, Engstrom, & Gilbert, 2002; Watkins et al., 2004). Seminal epidemiological research documents higher prevalence of psychological distress among justice-involved women than among women in the general population (Jordan, Schlenger, Fairbank, & Caddell, 1996; Teplin, Abram, & McClelland, 1996). For example, between 13% and 17% of women in jail and prison, respectively, are found to have major depressive disorder (Jordan et al., 1996; Teplin et al., 1996). These rates are two to three times higher than rates among women in the general population. High rates of PTSD are also found among female jail detainees; 22.3% of the women in one study met the lifetime diagnostic criteria for PTSD (Teplin et al., 1996). In a more recent national survey, it was found that 43% of all female jail inmates had a serious mental illness such as depression (28%), bipolar disorder (15%), or schizophrenia (4%), whereas 53% had a lifetime diagnosis of PTSD (Lynch et al., 2012). Incarcerated women with mental health problems are three times more likely to experience addiction than those without mental health problems (James & Glaze, 2006). Other data indicate that rates of co-occurring substance use and psychological distress (e.g., PTSD, any serious mental illness) range from 39% to 46% among female jail detainees (Lynch et al., 2012). The combination of substance use and psychological distress places justice-involved women at higher risk for other negative life events: unemployment, homelessness, increased criminal justice involvement, and victimization (Baillargeon et al., 2010; Engstrom, El-Bassel, Go, & Gilbert, 2008; James & Glaze, 2006; Marquart, Brewer, Simon, & Morse, 2001).

Victimization across the life span is highly prevalent among women involved in the justice system. Although estimates vary depending on the sample, measures, and method of data collection, data indicate that between 60% and 99% of women in the criminal justice system have experienced some form of physical, sexual and/or psychological victimization in their lives (McDaniels-Wilson & Belknap, 2008; Reichert, Adams, & Bostwick, 2010). Lawbreaking and incarceration have been linked empirically and theoretically to women’s experiences of victimization (Bloom, Owen, & Covington, 2003; Browne, Miller, & Maguin, 1999; McDaniels-Wilson & Belknap, 2008; Reichert et al., 2010; Salisbury & Voorhis, 2009; Tripodi & Pettus-Davis, 2013; Widom, 1995; Widom & Ames, 1994). For example, childhood abuse and neglect increase the likelihood of arrest as a juvenile by 59%, as an adult by 28%, and for a violent crime by 30%, regardless of gender (Widom & Ames, 1994). For females in particular, those who were abused or neglected in childhood are 73% more likely than a comparison group of women to be arrested for property, alcohol, drug, violent and misdemeanor charges such as disorderly conduct, curfew violations or loitering (Widom & Ames, 1994).

An increasing body of research across diverse populations of women suggests that the association between victimization and subsequent criminal justice involvement is influenced by substance use and psychological distress (Daly, 1992–1993; DeHart et al., 2013; Salisbury & Voorhis, 2009). However, research has yet to fully examine the complex relationships between various types of victimization (e.g., childhood abuse, adult intimate partner violence (IPV), and adult nonintimate partner violence [NIPV]), psychological distress, and substance use among justice-involved women specifically. In fact, most of the research to date on justice-involved women has been descriptive in nature, highlighting prevalence rather than illuminating the relationships among these factors (for notable exceptions, see Salisbury & Voorhis, 2009; Tripodi & Pettus-Davis, 2013). Given the ways in which substance use contributes to the risk of criminal justice involvement, especially in the context of the War on Drugs, it is particularly important to understand correlates of drug use in order to design effective services that can improve well-being and reduce drug use and its multifaceted consequences for women.

In order to address this gap, the present study sought to increase understanding of the associations between different types of victimization, psychological distress, lawbreaking activity and recent drug use among women in the criminal justice system. This study was guided by an adaptation of the Comprehensive Health Seeking and Coping Paradigm (CHSCP; Nyamathi, 1989); research guided by the CHSCP has made important contributions to understanding the mechanisms associated with behaviors such as substance use and HIV risk behaviors among adults experiencing poverty and other risks (Nyamathi, Flaskerud, & Leake, 1997; Nyamathi, Keenan, & Bayley, 1998; Nyamathi et al., 1999; Nyamathi, Leake, Keenan, & Gelberg, 2000; Nyamathi et al., 2012; Nyamathi, Stein, & Bayley, 2000; Nyamathi, Stein, & Swanson, 2000). Within this framework, women’s engagement in high-risk behavior, such as recent drug use, is conceptualized as a function of a multidimensional health-seeking process that is characterized as a transaction between the individual and her environment at multiple systemic levels (Nyamathi, 1989). In the present study, sociodemographic variables and indicators of personal resources (a theoretical construct that includes measures of psychological distress, self-esteem and impulsivity), victimization, and dynamic drug and crime factors are identified as potential domains that influence women’s engagement in recent drug use. The CHSCP was used to guide the selection of relevant behavioral domains and their operationalization (e.g., personal resources) and adapted to include domains that address the unique circumstances of justice-involved women (e.g., victimization, dynamic drug and crime factors).

Although theoretically guided, this research is exploratory and not driven by a priori hypotheses about relationships among domains; rather, this research seeks to provide the necessary evidence upon which to build well-defined, population-specific models of behavior. Thus, the present study addressed the following research question: what factors (i.e., sociodemographic characteristics, personal resources, type of victimization, and dynamic drug and crime factors) are associated with recent drug use among victimized women on probation and parole? In a population where drug use and victimization are highly prevalent, findings from the present study will allow practitioners and policymakers to more easily identify and assist women at higher risk for continued substance use, and potentially, further criminal justice involvement.

Participants and Procedures

The sample included 406 women on probation and/or parole in Jefferson County, Kentucky. Jefferson County is a large, urban area that includes Louisville. Recruitment methods included face-to-face recruitment at all probation and parole offices located within the county; direct mailings to women on probation and parole in Jefferson County; advertisements in the local newspaper; the website Craigslist; public access TV; fliers posted in a variety of public locations (e.g., bus stops, convenience stores, apartment complexes); community-based organizations; government agencies; health care facilities; and community outreach by study personnel.

To be eligible for participation, women had to meet the following criteria: a) be on probation and parole in Jefferson County; b) be at least 18 years of age; c) report that when they had sex they either had sex with men only or with both women and men (women who had been recently incarcerated were asked about the year before incarceration); and d) report lifetime experience of physical and/or sexual victimization as a child and/or an adult from a parent/caretaker, intimate partner, and/or nonintimate partner (e.g., stranger, acquaintance). Screening for eligibility was conducted by telephone (90%) and in person (10%). Eighty-one percent of the women screened were eligible to participate. Women who were screened reported learning about the study from the following sources (participants could identify more than one source): direct mail (33%); word of mouth (e.g., a probation officer, mother, friend; 33%); fliers posted in public locations (15%); community-based organization (11%); direct contact with study personnel (9%); and newspaper/radio/Internet (2%). The most common reasons for ineligibility were not being on probation or parole, no history of victimization, and reporting only female sexual partners.

Before the interview, the women were consented using the University of Louisville Institutional Review Board approved consent form; an National Institutes of Health (NIH) Certificate of Confidentiality was obtained and documented in the consent form. All interviews were administered face-to-face by trained female staff using audio computer-assisted interviewing (ACASI; Nova Research Company, 2003) on laptop computers; on average, interviews lasted 2 and a half to 3 hrs. Participants used headphones to listen as the survey items were read to them by the computer (questions and response options were also simultaneously displayed on the computer screen); participants entered their own responses directly into the computer. Questionnaires were password-protected and response data were encrypted so that unauthorized users were unable to view, export or modify collected data without the correct password. Participants were debriefed and compensated $35 for their time.

Independent Variables

Sociodemographic characteristics

Respondents’ age was provided in years. Two categories were used to describe the race/ethnicity of the participants: African American, multiracial, or other racial/ethnic background and White. Intimate partner status was assessed by three categories indicating whether a respondent reported being: single; married or cohabiting with a male sexual partner; or divorced, separated or widowed at the time of the interview. Three categories described educational attainment: less than a high school diploma/GED; high school diploma/GED; greater than a high school diploma/GED. Current employment status was dichotomous, working (1) or not working (0), and women were asked if they considered themselves homeless (yes = 1; no = 0).

Type of victimization

Victimization was operationalized by eight dichotomous variables (yes = 1; no = 0) reflecting whether a woman reported experiencing any of the following types of victimization: childhood psychological abuse, childhood physical abuse, childhood sexual abuse, psychological IPV, physical IPV, sexual IPV, physical NIPV, and sexual NIPV. This operationalization is consistent with a growing body of research that examines different types of childhood and adult victimization simultaneously (Alvarez et al., 2009; McGuigan & Middlemiss, 2005).

Variables comprised items adapted from the National Crime Victimization Survey, the Revised Conflict Tactics Scale and Tolman’s Psychological Maltreatment of Women Inventory (Straus, Hamby, Boney-McCoy, & Sugarman, 1996; Tjaden & Thoennes, 1998, 2000; Tolman, 1999; Tolman, 1989; a complete list of items is available from the first author) and have been used in prior research (Golder & Logan, 2010, 2011; Golder & Logan, 2014; Logan & Leukefeld, 2000; Logan, Walker, & Leukefeld, 2001). Childhood victimization assessed abuse by a parent and/or other caretaker when a woman was 18 or younger; IPV referred to perpetration of violence by individuals “like a boyfriend and/or husband” whereas NIPV was defined as victimization perpetrated by a stranger, acquaintance or relative (other than guardians/parents or spouses).

Psychological abuse in childhood and in the context of IPV captured a range of potentially psychologically abusive experiences (e.g., “insulted, shamed or humiliated you in front of others”; “withhold food from you as a punishment”). Physical childhood abuse assessed whether respondents had ever been physically hurt on purpose, beat up, or attacked with a weapon; physical IPV and NIPV assessed whether the corresponding person had ever: physically hurt her on purpose; caused her to have an accident; beat her up; used a knife, gun or some other thing (like a club or bat) to get something [from her]; and/or attacked her with a weapon. Childhood, IPV and NIPV sexual victimization were assessed by questions asking respondents if they had ever been forced or threatened to: do “sexual things other than sexual intercourse (e.g., petting, oral sex)”; “have sexual intercourse but it did not actually occur”; and/or “have sexual intercourse and it actually happened.”

Personal resources

Three areas were evaluated: psychological distress, self-esteem and impulsivity. Psychological distress was operationalized by two variables (general psychological distress; PTSD). General psychological distress was measured by the Global Severity Index (GSI) of the Brief Symptom Inventory (BSI; Derogatis, 1993), which yields a summary score that combines the nine symptom domains of the BSI. The individual is asked to describe her degree of distress for each psychological symptom during the past seven days; higher scores indicate higher levels of psychological distress/symptoms (range 0–4). Raw scores above .62 on the GSI indicate levels of reported psychological distress that exceed those of 84% of the national population of women and are considered to surpass the clinical threshold for presence of a mental health problem (Derogatis, 1993; Golder & Logan, 2010; Potter & Jenson, 2003). Alpha reliability for the present scale was .97.

PTSD was measured by the 49-item Posttraumatic Stress Diagnostic Scale (PDS; Foa, 1995; Foa, Cashman, Jaycox, & Perry, 1997). A single variable assessed whether or not (yes = 1; no = 0) the woman currently met the DSM–IV diagnostic criteria for PTSD. The PDS is found to be reliable, valid and have good diagnostic performance (Foa, 1995; Griffin, Uhlmansiek, Resick, & Mechanic, 2004; Powers, Gillihan, Rosenfield, Jerud, & Foa, 2012).

Self-esteem was measured by the Rosenberg Self-esteem Scale (Rosenberg, 1965). The scale has a possible range of 0 to 30, with higher scores reflecting lower self-esteem (alpha reliability in the present study = .87.). Lastly, impulsivity was measured by the Barratt Impulsiveness Scale-11, a widely used and well-validated personality measure (Patton, Stanford, & Barratt, 1995). The measure consists of 30 statements; representative items include “I don’t pay attention” and “I do things without thinking.” The response options range from 1 = “Rarely/Never” to 4 = “Almost Always/Always”; higher scores reflect higher levels of impulsivity (possible range = 30 to 120; alpha in the present study = .87).

Dynamic drug and crime involvement factors

Eight variables operationalized this construct. Previous studies have demonstrated that early substance use—even when such use does not meet criteria for a substance use disorder—is associated with an increased risk of adult illegal activity or criminality (Stenbacka & Stattin, 2007). In the current study, early drug use was operationalized as a dichotomous variable indicating whether or not a woman used any of 10 substances (i.e., marijuana, cocaine, crack, heroin, opiates other than heroin, hallucinogens, sedatives/tranquilizers/barbiturates, methamphetamine, club drugs and prescription drugs) before the age of 13 (yes = 1; no = 0); prior research has established the age of 13 as an appropriate cutpoint for early initiation of substance use (Grant & Dawson, 1998). Participants were asked whether they had ever been in alcohol or drug treatment (yes = 1; no = 0) and the total number of lifetime treatment episodes.

Correctional status was assessed by asking women to indicate whether they were on probation, parole or both. Questions adapted from the Addiction Severity Index (McLellan et al., 1992) assessed the number of days a woman reported being in a controlled environment and/or halfway house/recovery home in the past 12 months. Women reported the total number of days they had been incarcerated in the past 5 years.

Finally, two variables measured the number of lawbreaking activities a woman reported over her lifetime and over the past 12 months (possible range 0 to 8). Women were asked about their engagement in eight separate lawbreaking behaviors (e.g., “Purposely damaged or destroyed something that did not belong to you?”; “Knowingly received, bought or sold stolen goods”). Questions involving drug-involved lawbreaking were omitted to avoid overlap with the dependent variable.

Dependent Variable

Recent drug use

Because even a single episode of drug use can have significant implications for women on probation and parole, a dichotomous variable (yes = 1; no = 0) was used to assess whether a respondent reported use of any of the 10 substances identified above during the past 12 months.

Data Analysis

To address the primary research question, bivariate and multivariate analyses were conducted. Bivariate analyses were used to gain a better understanding of which independent variables best represented the theoretically specified domains under consideration. First, differences between women who reported recent drug use and those who did not report recent use were compared across the four domains of the CHSCP (sociodemographic characteristics, personal resources, lifetime victimization, dynamic crime and drug factors). Variables that evidenced a significant relationship with the dependent variable in this step were retained for inclusion in the multivariate analysis. In addition, bivariate correlations are presented for these variables, providing a measure of effect size (i.e., correlation coefficient) and a means for assessing for high multicollinearity. Finally, multivariate analysis using logistic regression was used to identify the best fitting model; tolerance statistics are presented for each variable in the model to allow for examination of multicollinearity. The overall analysis strategy is consistent with a theory-guided, model-building approach and the research question being addressed (Tabachnick & Fidell, 2001).

Descriptive Findings

Means or percentages, standard deviations, and range for all variables are reported in Table 1. Briefly, the women were on average 37 years of age, slightly more than half were White (50.5%), and 44.6% reported being single, not living with a male partner. Educational attainment varied; slightly more than 27% reported less than a GED or high school diploma whereas 32% reported more than a GED or high school diploma. Twenty-nine percent of the women reported working part- or full-time, and 34% considered themselves homeless. Select results for Personal Resources showed that 68.7% of the women in this sample had a level of general psychological distress that was greater than the cutoff for clinical significance and 48.5% of the women met the diagnostic criteria for PTSD.

Sample Characteristics and Between-Group Differences for Women Who Do and Do Not Report Recent Drug Use (Past 12 Months)

Approximately 70% of the women reported experiencing some form of physical and/or childhood sexual victimization, slightly more than 90% reported ever experiencing any physical and/or sexual IPV, and 72% reported physical and/or sexual NIPV sometime in their lives (data not reported in table). Rates of childhood victimization ranged from 38.7% (sexual) to 75.1% (psychological), whereas the prevalence of different types of IPV in the past 12 months ranged from 9.1% (sexual) to 49.0% (psychological). Slightly more than 7% of women reported sexual NIPV and 10.8% reported physical NIPV in the past 12 months. Approximately 42% of the women had engaged in drug use before the age of 13. Two-thirds of the participants had been in drug treatment sometime in their lives. The majority of women were on probation (75.6%); they had spent approximately 47 days in a controlled environment in the past 12 months and 286 days incarcerated in the past 5 years. On average, the women had engaged in almost four (3.78) different lawbreaking behaviors over their lives and about two (1.84) lawbreaking behaviors in the past year.

Approximately 46% of the women reported recent drug use. Regarding the types of substances that were used in the past 12 months, marijuana was the most common (27.9%) followed by opiates other than heroin (19.8%), cocaine (18.0%), sedatives/tranquilizers/barbiturates (17.0%), crack (14.8%), prescription drugs (13.1%), heroin (6.2%), club drugs (2.7%) and hallucinogens (.7%; data not reported in table).

Bivariate Analysis

Significant associations were found between 14 of the independent variables and the dependent variable. Women who used drugs were more likely to be younger than women who did not use drugs in the past 12 months and to report higher levels of general psychological distress (1.29 compared with 1.08) and PTSD (54.5% compared with 43.4%). With the exception of childhood psychological and physical victimization, there were statistically significant differences on all the measures of victimization between women who reported recent drug use and those who did not report recent drug use. Generally, victimization was associated with drug use; however, women who reported experiencing childhood sexual victimization were less likely to report recent drug use (33.2% compared with 43.4%). Regarding the dynamic drug and crime factors, women who used drugs recently were more likely to report initiating drug use at an early age (52.9% compared with 32.4%) and to have been in drug treatment in the past 12 months (46.0% compared with 35.2%). However, there were no significant differences in the number of times each group of women reported being in drug treatment. Finally, women who used drugs recently were more likely to be on probation (88.2%), to have spent fewer days incarcerated in the past 5 years (an average of 237.10 days in the past 5 years as compared with 327.94 days), and to have engaged in more lawbreaking behaviors across the two measured periods (4.18 and 2.42 ever and in the past 12 months, respectively). The correlation matrix is found in Table 2; there is no evidence of high multicollinearity among the variables (Allison, 2012).

Bivariate Correlation Matrix

Multivariate Analysis

The test of the final model, against a constant-only model was statistically reliable, chi-square (14, N = 399) = 102.001, p ≤ .01, indicating that the predictors, as a set, reliably distinguished between women who used drugs in the past 12 months and those who did not (−2LL = 446.237). The final model accounted for 30% (Nagelkerke R²: .303) of the variance in drug use over the past 12 months (Tabachnick & Fidell, 2001). The overall prediction success for this model was 71.8%, which was greater than the proportional by chance accuracy rate of 62.75%, thus supporting the utility of the model. More specifically, 77.5% of the women who reported not using drugs in the past 12 months and 65.2% of those who did use drugs were correctly predicted by the model.

Table 3 shows the regression coefficients, standard errors (SE), odds ratios (OR), 95% confidence intervals (CI), and tolerance for the final model. Five variables in the model reached the conventional level of significance: age, childhood sexual victimization, early drug use, correctional status, and lawbreaking (past 12 months). There was an inverse relationship (OR = .96; 95% CIs [.94, .99] between age and the dependent variable: older age was associated with slightly reduced odds of drug use in the past year. Similarly, there was an inverse relationship between childhood sexual victimization and recent drug use such that women who were sexually victimized as children were considerably less likely to report recent drug use than women who had not experienced this type of victimization (OR = .56; 95% CI = [.34, .91] Early drug use (OR = 2.63; 95% CI [1.60, 4.32] and engaging in more lawbreaking behaviors during the past 12 months (OR = 1.75; 95% CI [1.32, 2.31] were associated with recent drug use. Finally, correctional status was inversely related to the dependent variable (OR = .26; 95% CI [.13, .50] the odds of women on parole reporting drug use in the past 12 months were less than those of women on probation. Confirming the interpretation of the bivariate correlations, tolerance values were all well above .10; larger values (i.e., those closer to 1.0) are indicative of fewer problems with multicollinearity (Denis, 2011; Regression with SPSS Chapter 2: Regression Diagnostics).

Logistic Regression Model Predicting Recent Substance Use (Past 12 Months)

Drug use plays a major role in women’s involvement in the criminal justice system. Identifying and understanding factors associated with this high-risk behavior is essential in order to address it. This is the first known study to elucidate the associations between types of victimization, psychological distress and recent drug use among women on probation and parole. When controlling for confounding factors, the multivariate findings indicate that women who recently used drugs were younger, were not sexually victimized as children, began using drugs before they were 13 years of age, were on probation and had engaged in more recent lawbreaking. In terms of the CHSCP framework informing the study, these findings suggest that domains related to sociodemographic characteristics, victimization, drug use, illegal activity and correctional system status are salient correlates in recent drug use among women on probation and parole involved in this study. In order to further consider the relationships identified in the multivariate analysis, each of the significant predictors is considered in turn.

The association between age and substance use in the general population is well established. Research has identified an age-related pattern of drug and alcohol use that begins in adolescence, peaks in young adulthood, and then declines (Bachman, Wadsworth, O’Malley, Johnston, & Schulenberg, 1997; Chassin, Fora, & King, 2004; Chen & Kandel, 1995; van Lier, Vitaro, Barker, Koot, & Tremblay, 2009; Roettger, Swisher, Kuhl, & Chavez, 2011; Tobler & Komro, 2010). This pattern is evidenced in the most recent National Survey on Drug Use and Health: rates of substance dependence or abuse were higher among younger adults (age 18 to 25; 18.6%) than among youth (age 12 to 17; 6.9%) or adults 26 and older (age 26 and older; 6.3%; Substance Abuse and Mental Health Services Administration, 2012). Notwithstanding the general relationship between age and substance use, methodological advances have highlighted the heterogeneity found among people who use drugs and have contributed to a more nuanced understanding of different patterns of substance use across the life course (Guo et al., 2002; Hser, Anglin, Grella, Longshore, & Prendergast, 1997; Hser, Huang, Brecht, Li, & Evans, 2008; Kertesz et al., 2012). Research focused on adults with substance use histories suggests that the relationship between age and drug use behaviors is complex and does not necessarily decline with age (for discussion see Engstrom, Shibusawa, El-Bassel, & Gilbert, 2011). In fact, research with 60 incarcerated women in Kentucky found similar rates of recent substance use among older and younger women (Stanton, Walker, & Leukefeld, 2003).

Relatedly, there is a growing body of research that has identified a variety of distinct subgroups and trajectories among people who use drugs, with certain attention given to age at initiation (Chassin et al., 2004). In particular, early onset of drug use (as well as illegal activity) is associated with more severe and persistent drug use patterns (Hser et al., 2008; Hser, Longshore, & Anglin, 2007). Consistent with these findings, the present study found that odds of recent drug use were 2.45 times as large for women who used drugs before the age of 13 than for women who did not experience early drug use. The aforementioned results, together with research indicating that people with younger age of drug initiation and more involvement in lawbreaking may be less responsive to treatment (Grella & Lovinger, 2011), suggest that victimized women on probation and parole who initiated early drug use may be at particularly high risk for ongoing substance use and may require innovative outreach and intervention strategies. One approach that may hold promise for adaptation is the Comprehensive Gang Model. This model, which has been tested and refined in more than 20 communities in the U.S. since the 1990s, employs a comprehensive set of strategies that address both structural/community-level issues and individual-level approaches to address a wide range of needs (i.e., psychological distress, educational attainment) among gang-involved youth (Office of Juvenile Justice and Delinquency Prevention, 2010; Spergel, 1995). The specific strategies include: community mobilization involving citizens, former gang members (in the case of adaptation, these strategies would focus on justice-involved women rather than gang members), community groups and agencies in the coordination of community-based programming; opportunity for justice-involved youth to engage in education, training and employment programs; social intervention focused on linking justice-involved youth and their families to needed services; use of close supervision and monitoring by the criminal justice system, as well as community-based organization; and development and implementation of policies and procedures that capitalize on available and existing resources. Unfortunately, evidence-supported prevention programs such as the Comprehensive Gang Model have not been widely adopted in the U.S.

The relationship between child sexual abuse and later substance use among women is well documented (Gutierres & Puymbroeck, 2006; Ireland & Widom, 1994; Widom, Marmorstein, & White, 2006). As such, it is not unexpected that rates of child sexual victimization and recent drug use among this sample appear high. For example, the prevalence of child sexual abuse among a national sample of older adolescents indicates that among females 17 and younger, 26.6% and 5.5% report an incident of sexual abuse/assault by any type of perpetrator or family member, respectively (Finkelhor, Shattuck, Turner, & Hamby, 2014). In comparison, 38.7% of the women in the present sample reported experiencing childhood sexual victimization perpetrated by a parent and/or caretaker. Similarly, among all women 12 and over, only 12.5% report any past year use of illicit substances compared with 46% of the women in this study (Substance Abuse & Mental Health Services, 2009, 2010). Although the current study found high prevalence of childhood sexual victimization and recent drug use among participants, it also found an inverse relationship between childhood sexual victimization and recent drug use. Among women in this study, experiencing childhood sexual abuse was associated with a 44% reduction in odds of recent drug use. Thus, the relationship between these factors among this sample of highly victimized women appears complicated and may reflect the particular roles of current victimization, psychological distress and other factors in recent drug use. Future research may examine differences between the women who have been victimized sexually as children and those who have not in order to better understand these relationships. More specifically, research should determine if there are groups of women who have distinct profiles based on indicators of lifetime violence and how those profiles are associated with drug use and other indicators of psychosocial functioning.

Finally, two dynamic drug and crime factors were also significant predictors of recent drug use: correctional status and engaging in more recent lawbreaking behavior. The odds were greater that probationers, as compared with parolees, would report recent drug use (OR = .26; 95% CI [.14, .46] It may be that this finding indicates the presence of different trajectories of drug use (and/or lawbreaking behaviors) captured by the categorization/assignment of women to probation and parole. In fact, probation and parole represent different points on the criminal justice continuum (Center for Substance Abuse Treatment, 2005). As such, entry into, conditions of, and consequences for failure to comply with conditions may differ by type of supervision, thus influencing women’s engagement in drug-using behavior. For example, a common condition of both probation and parole is the prohibition of drug use. For parolees, who have experienced a period of incarceration, the risk of revocation and subsequent reincarceration that is associated with drug use may act as an inhibiting factor. In contrast, women on probation may or may not have experienced a sustained period of incarceration. Thus the possibility of incarceration may be less threatening and may exert less influence on their decision to use drugs. Future longitudinal research will provide the opportunity to examine this supposition and provide data on associations between correctional status/involvement and long-term patterns of substance use among this population.

Similarly, the association between recent lawbreaking and recent drug use appears to reflect the complicated, multidimensional relationship between lawbreaking and drug use. Specifically, the relationship between drug use and lawbreaking has been characterized as reciprocal in nature, such that there is a “multiplier” effect through which more engagement in one is associated with more engagement in the other (Prendergast, Huang, & Hser, 2008). Again, longitudinal research that provides an opportunity to track trajectories of both substance use and lawbreaking is needed to further clarify these complex relationships. Although there remains a need for future investigations, decades of research have consistently found that “substance abuse treatment offers the best strategy for interrupting the drug abuse/criminal justice cycle” (National Institute on Drug Abuse, 2012, p. 13), particularly in the context of current drug-related legislation and criminal penalties in the U.S. This large body of research, together with the findings from the current study, underscore the critical need to identify effective strategies to engage more justice-involved women in substance use treatment that is both trauma-informed and guided by principles of effective treatment for people involved in illegal activity and the criminal justice system (Marlowe, 2003; Miller & Najavits, 2012).

These findings should be considered in light of the limitations. Although it is estimated that the sample comprised approximately one-fifth of all women on probation and parole in Jefferson County at the time recruitment was initiated (Kentucky Department of Corrections, Division of Probation and Parole, personal communication, November 5, 2010), the generalizability is limited by nonrandom sampling. Only victimized women were included in this sample; thus, comparisons between nonvictimized and victimized women were not possible. However, given the prevalence of victimization among justice-involved women, the most relevant questions are not limited to differences between victimized and nonvictimized women. Rather, research that helps elucidate the heterogeneous nature of victimization (i.e., type of victimization as measured in the current study, nature of relationships involving violence, severity of violence and so on . . .) and the subsequent effects on behavior are also necessary to inform development of targeted interventions for this population. Notwithstanding, these results are not generalizable to all women on probation and parole. Women who reported only having sex with other women were excluded from participation. Intimate partner violence between same-gender female partners is an important and understudied issue. The dynamics of intimate partner violence between same-gender partners may be similar to and/or distinct from violence between opposite-gender partners; however, this empirical question/issue was outside the focus of the present study. Additionally, there was concern that inclusion of women who only had sex with other women would yield a subsample that was too small for meaningful analysis; examination of screening data indicated that only 4% of all women screened were excluded because they reported only same-gender sexual behavior. Further attention to this group of justice-involved women is an important priority for future research. Dichotomizing race/ethnicity does not allow for in-depth assessments of potential race- and ethnicity-related differences in recent drug use; future research would benefit from sampling and recruitment strategies that support adequate statistical power to conduct meaningful analysis in this area. Similarly, although the detection of any drug use is reason for violation and potential incarceration for individuals on probation and parole, measuring recent drug use without consideration of the quantity/frequency of use may not capture severity, or allow consideration of other indices of problematic use, including patterns and consequences of use. Further, the cross-sectional design limits the ability to determine temporal ordering of the variables; as such, inferences regarding directionality of associations with drug use should be viewed cautiously. This study relied on self-reports of sensitive information that may have yielded underreports of some behaviors; however, this risk was minimized with the use of ACASI technology, which enhances reporting of potentially sensitive information (Wolff & Shi, 2012).

Despite its limitations, this study makes important contributions to our understanding of recent drug use among women on probation and parole. The CHSCP offers a broad theoretical framework from which to examine complex relationships between sociodemographic characteristics, victimization, psychological distress, lawbreaking activity, justice system involvement, and drug use among this population. As such, it assists with informing multidimensional intervention strategies that may provide key avenues to reduced drug use among this population of women. Most notably, the findings of the present study highlight the potential value of services that can effectively address risks associated with younger age, early initiation of drug use, probation status, and ongoing involvement in lawbreaking activity among women on probation and parole with histories of victimization.

¹ The 2000 ADAM report marks the last time data were available for female arrestees. In the case of female and male arrestees, the number of men and women testing positive for these substances at arrest were almost identical, with 64% of men testing positive.

² A total of 7.6% of the respondents reported being Latina, Native American, Asian, multiracial or other racial/ethnic identity; the majority of these individuals (3.2%) identified as multiracial. Because the individual categories were too small for meaningful analysis, the groups were collapsed into a single group.

³ Prescription drug misuse was operationalized as ever using “prescription drugs that were not prescribed to you, in excess of what was prescribed for you, and/or for recreational purposes.”

⁴ The proportional by chance accuracy rate was computed by calculating the proportion of cases for each group (i.e., those cases that are correctly predicted and those that are incorrectly predicted) in the classification table at Step 0 and then squaring and summing the proportion of those cases for each group (0.463² + 0.537² = 0.502). Based on these calculations, the proportional by chance accuracy criteria is 62.75% (1.25 × 50.2% = 62.75%; Bayaga, 2010).

Allison, P. D. (2012). Logistic regression using SAS: Theory and application (2nd ed.). Cary, NC: SAS Institute Inc.

Alvarez, J., Pavao, J., Mack, K. P., Chow, J. M., Baumrind, N., & Kimerling, R. (2009). Lifetime interpersonal violence and self-reported chlamydia trachomatis diagnosis among California women. Journal of Women’s Health, 18, 57–63. 10.1089/jwh.2007.0665

Bachman, J. G., Wadsworth, K., O’Malley, P., Johnston, L., & Schulenberg, J. (1997). Smoking, drinking, and drug use in young adulthood: The impacts of new freedoms and new responsibilities. Mahwah, NJ: Erlbaum.

Baillargeon, J., Penn, J. V., Knight, K., Harzke, A. J., Baillargeon, G., & Becker, E. A. (2010). Risk of reincarceration among prisoners with co-occurring severe mental illness and substance use disorders. Administration and Policy in Mental Health, 37, 367–374. 10.1007/s10488-009-0252-9

Bayaga, A. (2010). Multinomial logistic regression: Usage and application in risk analysis. Journal of Applied Quantitative Methods, 5, 288–297.

Bloom, B., Owen, B., & Covington, S. (2003). Gender-responsive strategies: Research, practice, and guiding principles for women offenders. Washington, DC: National Institute of Corrections.

Browne, A., Miller, B., & Maguin, E. (1999). Prevalence and severity of lifetime physical and sexual victimization among incarcerated women. International Journal of Law and Psychiatry, 22, 301–322. 10.1016/S0160-2527(99)00011-4

Carson, E. A., & Sabol, W. (2012). Prisoners in 2011. Washington, DC: Bureau of Justice Statistics.

Center for Substance Abuse Treatment. (2005). Substance abuse treatment for adults in the criminal justice system: Chapter 10: Treatment for offenders under community supervision. Treatment Improvement Protocol (TIP) Series (Vol. No. 44).

Chassin, L., Fora, D. B., & King, K. M. (2004). Trajectories of alcohol and drug use and dependence from adolescence to adulthood: The effects of familial alcoholism and personality. Journal of Abnormal Psychology, 113, 483–498. 10.1037/0021-843X.113.4.483

Chen, K., & Kandel, D. B. (1995). The natural history of drug use from adolescence to the mid-thirties in a general population sample. Journal of Public Health, 85, 41–47. 10.2105/AJPH.85.1.41

Chilcoat, H. D., & Breslau, N. (1998a). Investigations of causal pathways between PTSD and drug use disorders. Addictive Behaviors, 23, 827–840. 10.1016/S0306-4603(98)00069-0

Chilcoat, H. D., & Breslau, N. (1998b). Posttraumatic stress disorder and drug disorders: Testing causal pathways. Archives of General Psychiatry, 55, 913–917. 10.1001/archpsyc.55.10.913

Conway, K. P., Compton, W., Stinson, F. S., & Grant, B. F. (2006). Lifetime comorbidity of DSM–IV mood and anxiety disorders and specific drug use disorders: Results from the National Epidemiologic Survey on Alcohol and Related Conditions. The Journal of Clinical Psychiatry, 67, 247–257. 10.4088/JCP.v67n0211

Cottler, L. B., Compton, W. M., III, Mager, D., Spitznagel, E. L., & Janca, A. (1992). Posttraumatic stress disorder among substance users from the general population. Journal of Psychiatry, 149, 664–670.

Daly, K. (1992–1993). Women’s pathways to felony court: Feminist theories of lawbreaking and problems of representation. Review of Law and Women’s Studies, 2, 11–52.

Dansky, B. S., Saladin, M. E., Brady, K. T., Kilpatrick, D. G., Resnick, H. S., & Addict, I. J. (1995). Prevalence of victimization and posttraumatic stress disorder among women with substance use disorders: Comparison of telephone and in-person assessment samples. The International Journal of the Addictions, 30, 1079–1099.

DeHart, D., Lynch, S., Belknap, J., Dass-Brailsford, P., & Green, B. (2013). Life history models of female offending: The roles of serious mental illness and trauma in women’s pathways to jail. Psychology of Women Quarterly, 38, 138–151. 10.1177/0361684313494357

Denis, D. J. (2011). Multiple linear regression using SPSS Part II. University of Montana. Retrieved from http://psychweb.psy.umt.edu/denis/datadecision/front/stat_II_2011/psyx_521_multiple_regression_part_II.pdf

Department of Justice. (2001). National Corrections Populations Reaches New High. Washington, DC: Bureau of Justice Statistics.

Derogatis, L. (1993). Brief Symptom Inventory: Administration, scoring, and procedures manual (3rd ed.). Minneapolis, MN: National Computer Systems, Inc.

Engstrom, M., El-Bassel, N., Go, H., & Gilbert, L. (2008). Childhood sexual abuse and intimate partner violence among women in methadone treatment: A direct or mediated relationship. Journal of Family Violence, 23, 605–617. 10.1007/s10896-008-9183-6

Engstrom, M., Shibusawa, T., El-Bassel, N., & Gilbert, L. (2011). Age and HIV sexual risk among women in methadone treatment. AIDS and Behavior, 15, 103–113. 10.1007/s10461-009-9625-z

Finkelhor, D., Shattuck, A., Turner, H. A., & Hamby, S. L. (2014). The lifetime prevalence of child sexual abuse and sexual assault assessed in late adolescence. The Journal of Adolescent Health, 55, 329–333. 10.1016/j.jadohealth.2013.12.026

Foa, E. B. (1995). Posttraumatic Stress Diagnostic Scale Manual. National Computer Systems Inc.

Foa, E. B., Cashman, L., Jaycox, L., & Perry, K. (1997). The validation of a self-report measure of posttraumatic stress disorder: The Posttraumatic Diagnostic Scale. Psychological Assessment, 9, 445–451. 10.1037/1040-3590.9.4.445

Glaze, L. (2002). Probation and Parole in the United States, 2001. Washington, DC: Bureau of Justice Statistics. 10.1037/e529812006-001

Golder, S., & Logan, TK. (2010). Lifetime victimization and psychological distress: Cluster profiles of out of treatment drug-involved women. Violence and Victims, 25, 62–83. 10.1891/0886-6708.25.1.62

Golder, S., & Logan, TK. (2011). Cumulative victimization, psychological distress, and high-risk behavior among substance-involved women. Violence and Victims, 26, 477–495. 10.1891/0886-6708.26.4.477

Golder, S., & Logan, TK. (2014). Violence, victimization, criminal justice involvement, and substance use among drug-involved men. Violence and Victims, 29, 53–72. 10.1891/0886-6708.VV-D-12-00040

Grant, B. F., & Dawson, D. A. (1998). Age of onset of drug use and its association with DSM–IV drug abuse and dependence: Results from the National Longitudinal Alcohol Epidemiologic Survey. Journal of Substance Abuse, 10, 163–173. 10.1016/S0899-3289(99)80131-X

Grella, C. E., & Lovinger, K. (2011). 30-year trajectories of heroin and other drug use among men and women sampled from methadone treatment in California. Drug and Alcohol Dependence, 118, 251–258. 10.1016/j.drugalcdep.2011.04.004

Griffin, M. G., Uhlmansiek, M. H., Resick, P. A., & Mechanic, M. B. (2004). Comparison of the posttraumatic stress disorder scale versus the clinician-administered posttraumatic stress disorder scale in domestic violence survivors. Journal of Traumatic Stress, 17, 497–503. 10.1007/s10960-004-5798-4

Guo, J., Chung, I.-J., Hill, K. G., Hawkins, J. D., Catalano, R. F., & Abbott, R. D. (2002). Developmental relationships between adolescent substance use and risky sexual behavior in young adulthood. The Journal of Adolescent Health, 31, 354–362. 10.1016/S1054-139X(02)00402-0

Gutierres, S., & Puymbroeck, C. V. (2006). Childhood and adult violence in the lives of women who misuse substances. Aggression and Violent Behavior, 11, 497–513. 10.1016/j.avb.2006.01.010

Hien, D. A., Jiang, H., Campbell, A. N., Hu, M.-C., Miele, G. M., Cohen, L. R., . . .Nunes, E. V. (2010). Do treatment improvements in PTSD severity affect substance use outcomes? A secondary analysis from a randomized clinical trial in NIDA’s Clinical Trials Network. The American Journal of Psychiatry, 167, 95–101. 10.1176/appi.ajp.2009.09091261

Hser, Y. I., Anglin, M. D., Grella, C., Longshore, D., & Prendergast, M. L. (1997). Drug treatment careers. A conceptual framework and existing research findings. Journal of Substance Abuse Treatment, 14, 543–558. 10.1016/S0740-5472(97)00016-0

Hser, Y. I., Huang, D., Brecht, M.-L., Li, L., & Evans, E. (2008). Contrasting trajectories of heroin, cocaine, and methamphetamine use. Journal of Addictive Diseases, 27, 13–21. 10.1080/10550880802122554

Hser, Y. I., Longshore, D., & Anglin, M. D. (2007). The life course perspective on drug use: A conceptual framework for understanding drug use trajectories. Evaluation Review, 31, 515–547. 10.1177/0193841X07307316

Ireland, T., & Widom, C. S. (1994). Childhood victimization and risk for alcohol and drug arrests. The International Journal of the Addictions, 29, 235–274.

James, D. (2004). Profile of jail inmates, 2002. Bureau of Justice Statistics.

James, D., & Glaze, L. (2006). Bureau of Justice Statistics Special Report, Mental Health Problems in Prison and Jail Inmates. Washington, DC: Bureau of Justice Statistics.

Jané-Llopis, E., & Matytsina, I. (2006). Mental health and alcohol, drugs and tobacco: A review of the comorbidity between mental disorders and the use of alcohol, tobacco and illicit drugs. Drug and Alcohol Review, 25, 515–536. 10.1080/09595230600944461

Jordan, B. D., Schlenger, W., Fairbank, J., & Caddell, J. (1996). Prevalence of psychiatric disorders among incarceratd women. Archives of General Psychiatry, 53, 513–519. 10.1001/archpsyc.1996.01830060057008

Kertesz, S. G., Khodneva, Y., Richman, J., Tucker, J. A., Safford, M. M., Jones, B., . . .Pletcher, M. J. (2012). Trajectories of drug use and mortality outcomes among adults followed over 18 years. Journal of General Internal Medicine, 27, 808–816. 10.1007/s11606-011-1975-3

Kessler, R. C., Chiu, W. T., Demler, O., Merikangas, K. R., & Walters, E. E. (2005). Prevalence, severity, and comorbidity of 12-month DSM–IV disorders in the National Comorbidity Survey Replication. Archives of General Psychiatry, 62, 617–627. 10.1001/archpsyc.62.6.617

Logan, TK., & Leukefeld, C. (2000). Violence and HIV risk behavior among male and female crack users. Journal of Drug Issues, 30, 261–282.

Logan, TK., Walker, R., & Leukefeld, C. (2001). Intimate partner and nonintimate violence history among drug-using, incarcerated men. International Journal of Offender Therapy and Comparative Criminology, 45, 228–243. 10.1177/0306624X01452008

Lynch, S., DeHart, D., Belknap, J., & Green, B. (2012). Pathways Project–Research Factsheet: Mental Health & Trauma among Women in Jails. Pocatello, ID: Idaho State University.

Marlowe, D. B. (2003). Integrating substance abuse treatment and criminal justice supervision. Addiction Science & Clinical Practice, 2, 4–14. 10.1151/spp03214

Marquart, J. W., Brewer, V. E., Simon, P., & Morse, E. V. (2001). Lifestyle factors among female prisoners with histories of psychiatric treatment. Journal of Criminal Justice, 29, 319–328. 10.1016/S0047-2352(01)00095-2

Maruschak, L., & Parks, E. (2012). Probation and parole in the United States, 2011. Washington, DC: Bureau of Justice Statistics.

McDaniels-Wilson, C., & Belknap, J. (2008). The extensive sexual violation and sexual abuse histories of incarcerated women. Violence Against Women, 14, 1090–1127. 10.1177/1077801208323160

McGuigan, W. M., & Middlemiss, W. (2005). Sexual abuse in childhood and interpersonal violence in adulthood: A cumulative impact on depressive symptoms in women. Journal of Interpersonal Violence, 20, 1271–1287. 10.1177/0886260505278107

McLellan, A. T., Kushner, H., Metzger, D., Peters, R., Smith, I., Grissom, G., . . .Argeriou, M. (1992). The fifth ed. of the Addiction Severity Index. Journal of Substance Abuse Treatment, 9, 199–213.

Miller, N. A., & Najavits, L. M. (2012). Creating trauma-informed correctional care: A balance of goals and environment. European Journal of Psychotraumatology, 3, 17246. 10.3402/ejpt.v3i0.17246

National Institute of Justice. (2003). 2000 arrestee drug abuse monitoring: Annual report. Washington, DC: U.S. Department of Justice.

National Institute on Drug Abuse. (2012). Principles of drug abuse treatment for criminal justice populations. Washington, DC: National Institute on Drug Abuse.

Nova Research Company. (2003). QDS Questionnaire Development System. Retrieved from http://www.novaresearch.com/index.cfm

Nyamathi, A. (1989). Comprehensive health seeking and coping paradigm. Journal of Advanced Nursing, 14, 281–290. 10.1111/j.1365-2648.1989.tb03415.x

Nyamathi, A., Flaskerud, J., & Leake, B. (1997). HIV-risk behaviors and mental health characteristics among homeless or drug-recovering women and their closest sources of social support. Nursing Research, 46, 133–137. 10.1097/00006199-199705000-00003

Nyamathi, A., Keenan, C., & Bayley, L. (1998). Differences in personal, cognitive, psychological and social factors associated with drug and alcohol use and nonuse by homeless womenResearch in Nursing & Health, 21, 525–532. 10.1002/(SICI)1098-240X(199812)21:6<525::AID-NUR6>3.0.CO;2-G

Nyamathi, A. M., Kington, R. S., Flaskerud, J., Lewis, C., Leake, B., & Gelberg, L. (1999). Two-year follow-up of AIDS education programs for impoverished women. Western Journal of Nursing Research, 21, 405–425. 10.1177/01939459922043956

Nyamathi, A., Leake, B., Keenan, C., & Gelberg, L. (2000). Type of social support among homeless women: Its impact on psychosocial resources, health and health behaviors, and use of health services. Nursing Research, 49, 318–326. 10.1097/00006199-200011000-00004

Nyamathi, A., Marlow, E., Zhang, S., Hall, E., Farabee, D., Marfisee, M., . . .Leake, B. (2012). Correlates of serious violent crime for recently released parolees with a history of homelessness. Violence and Victims, 27, 793–810. 10.1891/0886-6708.27.5.793

Nyamathi, A., Stein, J., & Bayley, L. (2000). Predictors of mental distress and poor physical health among homeless women. Psychology & Health, 15, 483–500. 10.1080/08870440008402008

Nyamathi, A. M., Stein, J. A., & Swanson, J. M. (2000). Personal, cognitive, behavioral, and demographic predictors of HIV testing and STDs in homeless women. Journal of Behavioral Medicine, 23, 123–147. 10.1023/A:1005461001094

Office of Juvenile Justice and Delinquency Prevention. (2010). Best practices to address community gang problems: OJJDP’s Comprehensive Gang Model (2nd ed.). National Gang Center.

Patton, J. H., Stanford, M. S., & Barratt, E. S. (1995). Factor structure of the Barratt impulsiveness scale. Journal of Clinical Psychology, 51, 768–774. 10.1002/1097-4679(199511)51:6<768::AID-JCLP2270510607>3.0.CO;2-1

Pew Center on the States. (2009). One in 31: The long reach of American Corrections. Washington, DC: The Pew Charitable Trusts.

Potter, C., & Jenson, J. (2003). Cluster profiles of multiple problem youth: Mental health problem symptoms, substance use, and delinquent conduct. Criminal Justice and Behavior, 30, 230–250. 10.1177/0093854802251007

Powers, M. B., Gillihan, S. J., Rosenfield, D., Jerud, A. B., & Foa, E. B. (2012). Reliability and validity of the PDS and PSS-I among participants with PTSD and alcohol dependence. Journal of Anxiety Disorders, 26, 617–623. 10.1016/j.janxdis.2012.02.013

Prendergast, M., Huang, D., & Hser, Y.-I. (2008). Patterns of crime and drug use trajectories in relation to treatment initiation and 5-year outcomes: An application of growth mixture modeling across three data sets. Evaluation Review, 32, 59–82. 10.1177/0193841X07308082

Regression with SPSS Chapter 2: Regression Diagnostics. Retrieved March 24, 2014, from http://www.ats.ucla.edu/stat/spss/webbooks/reg/chapter2/spssreg2.htm

Reichert, J., Adams, S., & Bostwick, L. (2010). Victimization and help-seeking behaviors among female prisoners in Illinois. Chicago, IL: Illinois Criminal Justice Information Authority. 10.1037/e586452010-001

Roettger, M. E., Swisher, R. R., Kuhl, D. C., & Chavez, J. (2011). Paternal incarceration and trajectories of marijuana and other illegal drug use from adolescence into young adulthood: Evidence from longitudinal panels of males and females in the United States. Addiction, 106, 121–132. 10.1111/j.1360-0443.2010.03110.x

Rosenberg, M. (1965). Society and the Adolescent Self-Image. Princeton, NJ: Princeton University Press.

Salisbury, E., & Voorhis, P. V. (2009). Gendered pathways: A quantitative investigation of women probationers’ paths to incarceration. Criminal Justice and Behavior, 36, 541–566. 10.1177/0093854809334076

Schiff, M., El-Bassel, N., Engstrom, M., & Gilbert, L. (2002). Psychological distress and intimate physical and sexual abuse among women in methadone maintenance treatment. The Social Service Review, 76, 302–320. 10.1086/339665

Spergel, I. (1995). Youth gang problem: A community approach. New York: Oxford University Press.

Stanton, M., Walker, R., & Leukefeld, C. (2003). Age differences in risk behavior among incarcerated substance-abusing women. Journal of Addictions Nursing, 14, 3–9. 10.1080/10884600305365

Stenbacka, M., & Stattin, H. (2007). Adolescent use of illicit drugs and adult offending: A Swedish longitudinal study. Drug and Alcohol Review, 26, 397–403. 10.1080/09595230701373875

Straus, M., Hamby, S. L., Boney-McCoy, S., & Sugarman, D. B. (1996). The revised Conflict Tactics Scales (CTS2): Development and preliminary psychometric data. Journal of Family Issues, 17, 283–316. 10.1177/019251396017003001

Substance Abuse and Mental Health Services. (2009, 2010). National Survey on Drug Use and Health, 2009 and 2010: Table 1.10B—Types of Illicit Drug Use in Lifetime, Past Year, and Past Month among Females Aged 12 or Older: Percentages, 2009 and 2010. Retrieved February 28, 2014, from http://oas.samhsa.gov/NSDUH/2k10NSDUH/tabs/Sect1peTabs1to46.htm#Tab1.10B

Substance Abuse and Mental Health Services Administration, National Survey on Drug Use and Health, 2009 and 2010. (2010). Table 5.4B—Substance Dependence or Abuse in the Past Year among Persons Aged 12 or Older, by Demographic Characteristics: Percentages, 2009 and 2010.

Substance Abuse and Mental Health Services Administration. NSDUH Series H-44. (2012). Results from the 2011 National Survey on Drug Use and Health: Summary of National Findings. Rockville, MD: Substance Abuse and Mental Health Services Administration.

Tabachnick, B., & Fidell, L. (2001). Using multivariate statistics (4th ed.). Needham Heights, MA: Allyn & Bacon.

Teplin, L. A., Abram, K. M., & McClelland, G. M. (1996). Prevalence of psychiatric disorders among incarcerated women. I. Pretrial jail detainees. Archives of General Psychiatry, 53, 505–512. 10.1001/archpsyc.1996.01830060047007

The Sentencing Project. (2012). Trends in U.S. Corrections. Washington, DC: The Sentencing Project.

Tjaden, P., & Thoennes, N. (1998). Stalking in America: Findings from the national violence against women survey. Washington, DC: National Institute of Justice, Centers for Disease Control and Prevention.

Tjaden, P., & Thoennes, N. (2000). Full report of the prevalence, incidence, and consequences of violence against women: Findings from the national violence against women survey. Washington, DC: U.S. Department of Justice. 10.1037/e514172006-001

Tobler, A. L., & Komro, K. A. (2010). Trajectories or parental monitoring and communication and effects on drug use among urban young adolescents. The Journal of Adolescent Health, 46, 560–568. 10.1016/j.jadohealth.2009.12.008

Tolman, R. M. (1989). The development of a measure of psychological maltreatment of women by their male partners. Violence and Victims, 4, 159–177.

Tolman, R. M. (1999). The validation of the psychological maltreatment of women inventory. Violence and Victims, 14, 25–37.

Tripodi, S. J., & Pettus-Davis, C. (2013). Histories of childhood victimization and subsequent mental health problems, substance use, and sexual victimization for a sample of incarcerated women in the US. International Journal of Law and Psychiatry, 36, 30–40. 10.1016/j.ijlp.2012.11.005

van Lier, P. A., Vitaro, F., Barker, E. D., Koot, H. M., & Tremblay, R. E. (2009). Developmental links between trajectories of physical violence, vandalism, theft, and alcohol-drug use from childhood to adolescence. Journal of Abnormal Child Psychology, 37, 481–492. 10.1007/s10802-008-9289-6

Watkins, K. E., Hunter, S. B., Wenzel, S. L., Tu, W., Paddock, S. M., Griffin, A., & Ebener, P. (2004). Prevalence and characteristics of clients with co-occurring disorders in outpatient substance abuse treatment. The American Journal of Drug and Alcohol Abuse, 30, 749–764. 10.1081/ADA-200037538

Widom, C. S. (1995). Victims of childhood sexual abuse–Later criminal consequences NIJ Research in Brief. Washington, DC: National Institute of Justice.

Widom, C. P., & Ames, M. A. (1994). Criminal consequences of childhood sexual victimization. Child Abuse & Neglect: The International Journal, 18, 303–318. 10.1016/0145-2134(94)90033-7

Widom, C. S., Marmorstein, N. R., & White, H. R. (2006). Childhood victimization and illicit drug use in middle adulthood. Psychology of Addictive Behavior, 20, 394–403.

Wolff, N., & Shi, J. (2012). Childhood and adult trauma experiences of incarcerated persons and their relationship to adult behavioral health problems and treatment. International Journal of Environmental Research and Public Health, 9, 1908–1926. 10.3390/ijerph9051908

Submitted: August 20, 2013 Revised: August 7, 2014 Accepted: August 19, 2014

Daily Associations Between PTSD, Drinking, and Self-Appraised Alcohol-Related Problems

Acknowledgement: This study was supported in part by a grant from NIH/NIAAA award (R21AA17130-01; PI: TLS) and by resources from the Center of Excellence in Substance Abuse Treatment and Education (CESATE; TLS), the VISN 20 and VISN 6 Mental Illness Research, Education, and Clinical Centers (MIRECC), the U.S. Department of Veterans Affairs Office of Academic Affiliations, Advanced Fellowship Program in Mental Illness Research and Treatment (SMW, MK, SSY, and KCB), and the VA Puget Sound Health Care System, Seattle, WA. The views expressed in this article are those of the authors and do not represent the views of the Department of Veteran Affairs or the United States government. Findings from the current study were disseminated previously as a poster presentation at the meeting of the International Society for Traumatic Stress Studies (Wilson, Krenek, Browne, Yard, & Simpson, 2015). The data used in the current study have yielded previously published articles (Browne, Wray, Stappenbeck, Krenek, & Simpson, 2016; Krenek, Lyons, & Simpson, 2016; Lehavot, Stappenbeck, Luterek, Kaysen, & Simpson, 2014; Stappenbeck, et al., 2015; Simpson, Stappenbeck, Luterek, Lehavot, & Kaysen, 2014).

Alcohol use disorder (AUD) and posttraumatic stress disorder (PTSD) often co-occur. Among individuals with PTSD, AUD is quite common, with estimates of lifetime prevalence ranging from 22–75% and with population-level estimates at 42% (Grant et al., 2015; Jacobsen, Southwick, & Kosten, 2001; Pietrzak, Goldstein, Southwick, & Grant, 2011). Similarly, individuals with lifetime AUD are at increased risk for lifetime PTSD (Goldstein et al., 2016). Much of the impairment associated with comorbid PTSD and AUD manifests with regard to difficulties in relationships, occupational impairment, legal problems, and health concerns (Blanco et al., 2013; Drapkin et al., 2011). For example, both PTSD and AUD increase the risk for health-related problems and all-cause mortality (Boscarino & Figley, 2009; Chwastiak, Rosenheck, Desai, & Kazis, 2010; Pacella, Hruska, & Delahanty, 2013), and their comorbidity is associated with greater physical health concerns and lower quality of life than seen for those with one or the other disorder (Evren et al., 2011; Kaysen et al., 2008). Although PTSD and AUD often exacerbate each other (Marshall et al., 2012; Oslin, Cary, Slaymaker, Colleran, & Blow, 2009; Wolitzky-Taylor, Bobova, Zinbarg, Mineka, & Craske, 2012), limited research has addressed the interplay between PTSD symptoms, drinking, and alcohol-related problems.

There is evidence that PTSD symptomatology is associated with AUD severity in treatment-seeking populations, and moreover this effect is not explained by quantity of alcohol consumed. In a sample of treatment-seeking U.S. veterans, those with comorbid PTSD and AUD were compared with individuals with AUD alone (Fuehrlein, Ralevski, O’Brien, Jane, Arias, & Petrakis, 2014). Although they had significantly fewer drinks per day and fewer heavy drinking days, participants with comorbid AUD/PTSD presented with significantly higher alcohol dependence scores (Fuehrlein et al., 2014). As this finding suggests, those with comorbid PTSD have AUD symptoms that may not be completely contingent upon quantity of alcohol consumed. In another study of treatment-seeking veterans, despite having fewer years drinking and no difference in current drinking quantity/frequency, those with comorbid PTSD/AUD reported greater alcohol dependence (AD; Petrakis et al., 2006). Taken together, these findings suggest that among individuals with AUD, comorbid PTSD may exacerbate alcohol’s impact on alcohol-related problems.

Building upon cross-sectional evidence of elevated alcohol problems in those with comorbid PTSD/AUD, Gaher and colleagues (2014) used experience sampling to examine within-person and between-person relationships between PTSD, drinking, and self-rated drinking problems. To examine daily and overall relationships with alcohol-related problems, it was necessary to statistically disaggregate effects of within-person PTSD and drinking (daily levels) from between-person PTSD and drinking (overall levels). In the study, within-person PTSD and drinking were assessed multiple times per day, and between-person PTSD was assessed by averaging each participant’s responses over the 2 weeks of monitoring. The authors showed that after controlling for the effects of drinking, both within-person and between-person PTSD independently affect self-ratings of alcohol-related problems. Within-person daily PTSD symptoms were associated with increased alcohol-related problems reported later the same day after controlling for quantity of alcohol consumed. Similarly, those with more severe between-person PTSD tended to report more alcohol-related problems after controlling for alcohol consumption.

Despite evidence that PTSD affects alcohol-related problems after controlling for drinking quantity, it remains unknown whether PTSD moderates the relationship between drinking amount and perceived alcohol-related problems. If PTSD severity does have a moderating effect on the relationship between drinking quantity and self-ratings of alcohol-related problems in treatment-seeking populations with comorbid PTSD/AUD, this could potentially affect PTSD/AUD treatment strategies. According to prevailing theoretical orientations to AUD treatment, behavior change is made possible in part through awareness and accurate assessment of alcohol-related problems (Donovan, 2003; Prochaska & Velicer, 1997). For individuals entering alcohol treatment, the more accurate they are in their self-appraisal of their pretreatment alcohol-related problems, the more positive their treatment outcome (Sawayama et al., 2012). Given that those with unremitting PTSD fare worse in AUD treatment outcome (Read, Brown, & Kahler, 2004), it is possible that PTSD contributes to this disparity by either exacerbating alcohol-related problems or disrupting accurate self-rating of alcohol-related problems.

While the aforementioned evidence clarifies that PTSD symptomatology adds unique variance to daily self-ratings of alcohol-related problems, we currently do not know whether PTSD symptoms moderate the cross-sectional association between drinking behavior and self-reports of alcohol-related problems. Varying levels of daily and overall PTSD symptom severity may differentially interact with heavy and lighter alcohol consumption to inform alcohol-related problem ratings. Both within person daily levels of PTSD and between person overall levels of PTSD could interact with daily drinking amounts. Thus, the present study used pretreatment daily monitoring data from an experimental treatment study (Simpson et al., 2014) to test the hypothesis that the association between daily drinking and same-day alcohol-related problems varies as a function of within-person PTSD (same-day symptoms) and between-person PTSD (overall severity).

Participants

Study participants were adults with concurrent diagnoses of AD and PTSD who indicated a desire to decrease alcohol use as part of a larger brief intervention study registered through ClinicalTrials.gov (Protocol #: NCT00760994). Briefly, study inclusion criteria were the following: (a) at least 18 years of age, (b) current AD diagnosis as defined by the Diagnostic and Statistical Manual of Mental Disorders, fourth edition, text revision (DSM–IV–TR; American Psychiatric Association, 2000), (c) alcohol use within the past 2 weeks, (d) current DSM–IV PTSD diagnosis, (e) capacity to provide informed consent, and (f) telephone access. Exclusion criteria were as follows: (a) history of delirium tremens or seizures, (b) opiate use or chronic treatment with opioid-containing medications during the past month, (c) Antabuse or naltrexone treatment, (d) alcohol withdrawal symptoms at initial consent, (e) acute suicidality/homicidality with intent/plan, or (f) psychosis. See Simpson et al. (2014), for further information on study participants and full inclusion/exclusion criteria. Data were excluded from the present analysis for participants with less than 50% adherence to daily monitoring to minimize nonresponse bias.

Ninety-two participants met the study entry criteria, and the final sample consisted of those who had at least 50% adherence on the daily monitoring protocol. Three participants were excluded because they did not complete monitoring days, and an additional three were excluded because they had less than 50% adherence on daily monitoring, yielding a final sample size of 86. Participants were on average 44.7 years of age (SD = 11.0), and ages ranged from 21 to 63. Ethnic group breakdown was as follows: 43.0% African American, 39.5% European American, 4.7% Hispanic/Latino, 4.7% American Indian, 1.2% Asian American, and 7.0% other ethnicity or missing. Almost half of the study sample was women (49.0%) and just over a quarter were veterans (25.6%). Only 13.1% were currently married, nearly a quarter were homeless (23.3%), and 12.8% were employed at least part-time, 10.5% were students, 45.3% were retired or disabled, and 31.4% were unemployed.

Procedure

Participants were recruited through newspaper advertisements and flyers. Participants completed an initial phone screen and then came into the lab where they provided written informed consent, underwent further screening for study inclusion, and a baseline assessment consisting of interview and self-report measures. They were compensated $30 for completing the baseline assessment. Participants received instruction on the telephone daily Interactive Voice Response (IVR) protocol. All study procedures were approved by the VA Puget Sound Health Care System Human Subjects Division Internal Review Board.

Participants completed self-report measures daily by calling a designated telephone number and answering prompts with IVR. Compliance was tracked automatically, and participants who did not call were personally contacted within two business days to collect data (10% of calls were collected verbally). Compensation was $1 for every completed day of monitoring, with a $10 bonus for seven consecutive monitoring days or a $7 bonus for six consecutive days. The time interval of IVR data collection was targeted for the 7 days before receipt of a brief intervention but this pretreatment baseline period ranged from 6 to 20 days because of scheduling difficulties. For additional information regarding procedures, see Simpson et al. (2014).

For the 86 participants who completed IVR monitoring, available observations ranged from 4 to 16 days (M = 7.3, SD = 2.6), with 659 total possible observations for all participants. The final IVR dataset, including all entries with available outcome data, contained 620 daily observations.

Measures

Screening assessment

Inclusion and exclusion criteria were assessed with the following measures: Hamilton Depression Inventory to assess suicidality (Hamilton, 1960); Structured Clinical Interview for DSM–IV Axis I Disorders to assess alcohol dependence, opiate use, and psychotic disorder (First, Gibbon, Spitzer, & Williams, 1995); PTSD Symptom Scale-Interview Version (PSS-I) to ascertain past-month PTSD diagnosis (Foa, Riggs, Dancu, & Rothbaum, 1993); and Form-42 (adapted from Form-90; Miller & Del Boca, 1994) to ensure alcohol use within the past 2 weeks. For additional information regarding these measures, see Simpson et al., 2014.

Baseline demographics

Participants were asked to identify age, gender, and veteran status as demographic covariates, and gender and veteran status were coded dichotomously (veteran status, 0 = nonveteran, 1 = veteran; gender, 0 = male, 1 = female).

Baseline PTSD

Participants were rated on baseline PTSD symptoms using the PSS-I, which is a 20- to 30-min, 17-item semistructured interview assessing DSM–IV symptoms of PTSD. The interviewer rates the severity of each symptom on a scale ranging from 0 (not at all) to 3 (5 or more times per week/very much), which yields a total score ranging from 0 to 51. The PSS-I shows excellent validity when compared to the Clinician-Administered PTSD Scale (Foa & Tolin, 2000). The PSS-I had good internal reliability in the current sample (α = .81).

Daily IVR self-report

Participants self-reported alcohol intake, PTSD symptoms, and alcohol-related problems daily using IVR telephone monitoring. Day of week (weekday = 0, weekend day = 1) was also recorded. Use of daily self-report IVR minimizes recall bias and underreporting of drinks per day (Searles, Helzer, Rose, & Badger, 2002).

IVR drinking

Participants were queried regarding the number of standard drinks consumed the day prior (beer, wine, and liquor, respectively). The number of each type of standard drink consumed each day was summed to yield a total drinks per day variable. Abnormally high values on this measure were verified verbally with participants. This methodology has been previously validated against retrospective self-report (Krenek, Lyons, & Simpson, 2016).

IVR PTSD symptoms

As described in Simpson et al., 2014, items assessing daily PTSD symptomatology were adapted from the PCL-C (King, Leskin, King, & Weathers, 1998). Items included three re-experiencing symptoms (intrusive thoughts, nightmares, and upset because of reminders), two avoidance symptoms (avoidance of thoughts and feelings associated with event, avoidance of trauma reminders), three emotional numbing symptoms (loss of interest, feeling detached, and emotionally numb), and four hyperarousal symptoms (concentration problems, alertness/vigilance, exaggerated startle, and anger/irritability). Items were chosen based on those that load most strongly in factor analytic studies of PTSD (e.g., King et al., 1998; Krause, Kaltman, Goodman, & Dutton, 2007; Palmieri, Weathers, Difede, & King, 2007) and that were likely to vary daily. Participants indicated how bothered they were on the previous day by each symptom on 9-point scales ranging from 0 = not at all to 8 = all the time. Items were averaged to create a daily IVR PTSD severity score (α = .94).

IVR alcohol-related problems

Self-appraised alcohol-related problems were assessed with a single item, “Yesterday, to what extent did you experience any negative consequences or problems related to your drinking?” Participants indicated a response on a 9-point scale ranging from 0 = none at all to 8 = worse ever. This single-item method of assessing daily alcohol-related problems was adapted from Searles and colleagues (2000).

Data Analytic Approach

Missingness was assessed for any associations with key variables at the between-subjects level (gender, veteran status, ethnicity, and baseline PTSD severity), and the within-subject level (weekend/weekday). Given nesting within the IVR data (daily self-reports nested within person), multilevel modeling (MLM) was used to analyze the data. The continuous outcome variable, daily alcohol-related problems, was zero-inflated (48.2% of responses were 0) and overdispersed (M = 2.16, SD = 2.53), corresponding to a zero-inflated gamma distribution. As such, we modeled daily alcohol-related problems via a two-part analysis. First, we used logistic MLM to model the odds of reporting any alcohol-related problems on a given day as a function of whether or not alcoholic drinks were consumed that day and PTSD symptom severity. Then, we used gamma-adjusted MLM to model nonzero alcohol problem ratings as a function of number of alcoholic drinks consumed and PTSD symptom severity.

Given that PTSD symptom severity data were collected longitudinally over a period of several monitoring days, it was possible to disaggregate within-person and between-person effects of PTSD symptom severity on alcohol problems (Curran & Bauer, 2011). To do so, we mean-standardized (i.e., z-scored) PTSD severity to statistically partial out effects of within-person daily PTSD symptoms opposed to overall, between-person PTSD symptoms over the entire monitoring period. Within-person PTSD severity was person-mean standardized (PMS) to capture the extent to which PTSD symptoms deviated from each participant’s personal mean on each day of monitoring. In other words, PMS PTSD reflects how mild/severe the participants’ PTSD symptoms were each day compared with their own personal average. We calculated between-person PTSD by grand-mean standardizing (GMS) each person’s overall PTSD scores. GMS PTSD, therefore, quantified the relative severity of each participant’s PTSD over the entire IVR monitoring period compared with others in the sample.

In each of the two models, we modeled the main effects of drinking and PTSD symptom severity on alcohol problems, and their interactions. In the logistic analysis, we examined the cross-level interaction between GMS PTSD (between-person differences in PTSD severity) and number of drinks consumed that day (Number of Drinks × Overall PTSD). We also examined the within-level interaction between PMS PTSD (within-person differences in PTSD) and number of drinks consumed that day (Number of Drinks × Daily PTSD). In the gamma model, we examined the interaction between GMS PTSD and number of drinks consumed that day (Number of Drinks × Overall PTSD) as well as the interaction between PMS PTSD and number of drinks consumed that day (Number of Drinks × Daily PTSD). In addition to these variables, veteran status, gender, age, time (days since beginning IVR monitoring), and weekend day versus weekday were covaried in both models. Logistic and gamma-adjusted MLM were conducted using PROC GLIMMIX, available in SAS (Version 9.4).

Prior preliminary analyses did not yield any significant differences between those included (n = 86) and excluded (n = 6) from the analyses with regards to baseline demographic characteristics or baseline PTSD symptomatology (see Simpson et al., 2014). Regarding missing IVR observations, there were no statistically significant patterns of missingness with regard to gender, veteran status, ethnicity, baseline PTSD severity, or weekend/weekday. At baseline, the average PSS-I score was 29.5 (SD = 9.2). See Table 1 for within-person and between-person variable mean values and SDs for the entire monitoring period. As a preliminary step, we explored within-person and between-person variability in IVR PTSD symptoms over the monitoring period by calculating an intraclass correlation (ICC) for the normally distributed IVR PTSD variable (Hoffman & Stawski, 2009).The ICC (0.695) indicated that 69.5% of the variability in PTSD symptoms was at the between-person level and 30.5% of the variability was at the within-person level.

Between-Subject and Within-Subject Descriptive Statistics, N = 620 Observations Among N = 86 Participants

Results from the logistic and gamma multilevel models are displayed in Table 2. According to the logistic model, the odds of reporting any alcohol-related problems increased when any alcohol was consumed, and individuals with higher between-person GMS PTSD severity were more likely to report having had an alcohol-related problem. Neither of the interactions between PTSD symptom level (PMS and GMS) and number of drinks consumed were significant.

Multilevel Models of Daily Alcohol-Related Problems

Turning to the gamma model, the main effects of number of drinks consumed, daily PMS PTSD symptom scores, and between-person GMS PTSD severity were each positively associated with number of alcohol-related problems reported. These main effects were qualified by significant interactions. According to the Number of Drinks × Daily PTSD interaction, the effect of number of drinks on alcohol-related problems varied as a function of daily PTSD level. At low levels of daily PTSD (1 SD below the mean) the multiplicative effect of each additional drink on self-reported problems was 2.03 (p < .001), whereas it was 1.87 (p < .001) at high levels of daily PTSD (1 SD above the mean). Contrasts in number of alcohol-related problems reported by daily PTSD level were most evident when few drinks were consumed. On days when more drinks consumed, there was less contrast in alcohol-related problems reported by daily PTSD level. There was a small effect of within-person daily PTSD on ratings of alcohol-related problems on nondrinking days (Cohen’s d = 0.25, p = .024) and on light drinking days (defined as three drinks consumed; Cohen’s d = 0.24, p = .045). There was no significant effect of within-person PTSD on self-rated problems for heavy drinking days (defined as nine drinks consumed; Cohen’s d = 0.05, p = .68).

The moderating effect of PTSD was even more pronounced in the Number of Drinks × Overall PTSD interaction (see Figure 1). According to that, the effect of number of drinks on alcohol-related problems varied as a function of overall PTSD severity level. As seen in Figure 1, at low levels of between-person PTSD (1 SD below the mean), the multiplicative effect of each additional drink on self-reported problems was 2.29 (p < .001), whereas it was 1.65 (p < .001) at high levels of PTSD (1 SD above the mean). Thus, the association between number of drinks consumed and self-rated alcohol-related problems was weaker for those with more severe between-person PTSD. The greatest contrasts in reported alcohol-related problems by individual differences in PTSD symptom level occurred on nondrinking days (Cohen’s d = 0.45, p < .001) and light-drinking days (Cohen’s d = 0.30, p = .007). However, as the number of drinks increased, there was less of an effect of between-person differences in PTSD symptom level on number of alcohol-related problems reported. At high levels of drinking (i.e., nine drinks), there was a trend toward a small inverse association between PTSD severity and alcohol-related problems, but this effect was not significant (Cohen’s d = 0.16, p = .13). Generally, at higher levels of between-person PTSD severity, there was a weaker association between quantity of alcohol consumed and self-rated alcohol-related problems.

Figure 1. Plot of modeled effects on alcohol-related problems by number of drinks consumed and between-person overall PTSD severity. Low and high PTSD scores were derived by calculating 1 SD offsets from the mean. Effect sizes and corresponding p values reflect contrasts between mean PTSD symptom levels and 1 SD offsets. PTSD = posttraumatic stress disorder; GMS = grand-mean standardized.

This study examined the daily effects of PTSD severity and alcohol consumption on alcohol-related problems among treatment-seeking men and women with co-occurring PTSD and alcohol dependence. Consistent with prior research (Gaher et al., 2014), we found that daily variability in PTSD severity, overall between-person PTSD severity, and alcohol consumption were all broadly predictive of alcohol-related problems. Our research extends these findings by providing evidence for moderating effects of both overall PTSD severity and daily PTSD symptom variability on self-appraised alcohol-related problems. We found that on nondrinking days and moderate drinking days (three drinks consumed), greater daily within-person PTSD symptom severity was associated with greater alcohol-related problems. Additionally, compared with participants with lower overall between-person PTSD scores, participants with higher overall PTSD severity reported higher ratings of alcohol-related problems on nondrinking and moderate-drinking days. This pattern was not apparent, however, on heavier drinking days, when neither within- nor between-person PTSD were associated with degree of alcohol-related problems. Given the strength of the moderating effect of between-person PTSD, these results suggest that for those with more severe PTSD there is a less contingent association between how much they drink on a given day and the extent to which they report negative consequences from drinking on nondrinking and moderate-drinking days.

These results may help to shed light on the prior research indicating that treatment-seeking individuals with comorbid PTSD and AUD, compared with those with AUD only, report greater alcohol dependence severity despite comparable or lower levels of consumption (Fuehrlein et al., 2014; Petrakis et al., 2006). In the present study, the weaker association between consumption and alcohol-related problems among those with higher levels of PTSD may be due in part to the bidirectional relationship between PTSD and negative coping strategies (Read, Griffin, Wardell, & Ouimette, 2014), suggesting that those with higher levels of PTSD may engage in negative coping strategies that lead to negative alcohol-related experiences regardless of quantity of alcohol consumed. Similarly, individuals with high levels of PTSD may experience negative effects from even small amounts of alcohol given positive associations between impulsivity/emotion dysregulation, and problematic alcohol use among those with PTSD (Schaumberg et al., 2015; Tripp & McDevitt-Murphy, 2015). It is noteworthy that among individuals with high levels of PTSD, the mean score for alcohol-related problems fell in the middle of the scale, suggesting that the weaker association between alcohol use and problems cannot be explained by a measurement ceiling effect.

It is also possible that alcohol craving or abstinence effects (i.e., withdrawal, hangover) were particularly detrimental to participants with higher PTSD on days with low or no alcohol use. Prior research has shown that the relationship between PTSD and alcohol-related consequences is mediated by alcohol craving (Tripp et al., 2015), and that exposure to trauma cues is linked to increased alcohol craving (Coffey et al., 2002). If those with higher PTSD have higher levels of craving on low- or nondrinking days, this could explain their self-rating of alcohol problems during days when they are not drinking as heavily. Indeed, both overall PTSD severity and daily increases in PTSD symptoms were associated with moderately high ratings of alcohol-related problems on nondrinking days. Additionally, experiences of hangover or alcohol withdrawal may be more aversive for those with severe PTSD, or conversely may temporarily exacerbate within-person PTSD symptoms.

Additionally, individuals with severe PTSD often have cognitive biases that may affect their accurate appraisal of alcohol-related problems. A core component of the disorder is a tendency to view experiences in a negative light and appraise situations as inherently dangerous (Vythilingam et al., 2007). Additionally, those with PTSD show memory bias toward trauma- and threat-related stimuli (Paunovic, Lundh, & Öst, 2002). Thus, a person with higher PTSD symptoms may recall a relatively minor incident that occurred after a single drink (e.g., an argument) as more distressing or problematic than someone with lower levels of PTSD. Similarly, this person may find it difficult to distinguish or attend accurately to the nuances between distinct incidences of varying importance or extremity, thus remembering them all with a similar valence. Additionally, given some particularly high responses with regards to drinking quantity, it is possible that memory bias also affected reporting of drinking patterns.

An interesting find was that the results indicated that between-person differences in PTSD more strongly moderated the association between alcohol consumption and problems than intraperson variability in PTSD. One possible explanation for this result is that the individual variability in PTSD severity was low. However, it is also possible that compared with daily PTSD, overall PTSD severity is a more sensitive marker for the negative impact of PTSD on functioning and perceived effects of alcohol use. This second explanation makes sense in the context of evidence for latent classes of PTSD that differ based upon overall, chronic symptomatology (e.g., Cloitre, Garvert, Weiss, Carlson, & Bryant, 2014; Galatzer-Levy, Nickerson, Litz, & Marmar, 2013).

Amplified alcohol-related problems on nondrinking and moderate-drinking days among those with more severe PTSD could be an important factor in treatment for individuals with comorbid PTSD and AUD. For example, individuals with more severe PTSD who overestimate alcohol-related problems may feel greater motivation to seek out and adhere to treatment. On the other hand, this overestimation may lead to lower self-efficacy for treatment success, which could partially explain the poor treatment outcomes often seen for this group (Saxon & Simpson, 2015). Additionally, this pattern of highly rated drinking problems on low-drinking days could be symptomatic of attentional biases that interfere with the accurate self-assessment that contributes to successful PTSD and AUD treatment (Donovan, 2003; Prochaska & Velicer, 1997). In other words, if a patient’s attentional bias causes them to consistently overestimate the role of alcohol in their life problems, this may interfere with their ability to make and achieve behavior change goals. Ultimately, treatment to address both PTSD and AUD may require interventions that support accurate assessment of drinking consequences either to increase motivation to engage in treatment or to clarify the impact of treatment as it (and its corresponding gains) occurs, or both.

These results should be viewed in the context of some study limitations. First, findings from this study were specific to a treatment-seeking sample, who may already be making efforts to reduce alcohol consumption. Thus, findings may not generalize to individuals with comorbid PTSD/AUD who are not seeking treatment. Additionally, all measures were self-report, which increases the likelihood of misinterpretation and other biases. Also, alcohol-related problems were assessed with a single item that asked about participants’ subjective experience of problems rather than the number of problems or specific types of problems. Given that this measure did not specify whether problems were because of active use or because of withdrawal, a more detailed measure of alcohol-related problems should be included in future studies. Measures also did not include consumption of other illicit substances, such as marijuana, opiates, cocaine, and methamphetamines. In addition, though only 1-day prior, retrospective reporting of PTSD symptoms, alcohol use, and problems, may have resulted in recall bias. However, questioning respondents at the end of the day would have increased the risk of missing consumption data and alcohol-related problems occurring later in the evening or early morning. In addition, we were only able to monitor participants for 1 week before receipt of a brief intervention; future research should use a longer monitoring period to better assess within-person changes and to determine whether these findings are maintained across time. Furthermore, given our sample of treatment seeking individuals with comorbid PTSD and alcohol dependence, results must be generalized as such. However, a strength of the study is that the sample included both veterans and civilians with almost equal distribution across gender. Finally, future research should include comparison groups with other psychological disorders or alcohol dependence only to parse out the influence of PTSD-specific symptomatology.

Despite these limitations, the results of the present study could provide guidance for future investigations on this important topic, including examination of potential mediators. For instance, cognitive symptoms such as catastrophizing or negative rumination may explain the weaker association between consumption and problems for those with more severe PTSD, as discussed above. Alternatively, negative coping processes (e.g., avoidance and anger reactivity) that are common components of PTSD may be contributing to this effect by compounding or prolonging alcohol’s effects. Additionally, gender may also moderate the effect of alcohol consumption and PTSD symptoms on self-ratings of alcohol-related problems. Further research should include consideration of coping mechanisms to increase our understanding of alcohol use and treatment in the context of PTSD.

¹ Previously published findings from this dataset showed modest lagged effects of posttraumatic stress disorder (PTSD) on next-day number of drinks (Simpson et al., 2014). In the current study, additional models not presented in current results demonstrated that there were no lagged effects of PTSD on alcohol-related problems. We ran models of lagged effects of PTSD on alcohol-related problems (controlling for same-day PTSD) and found no lagged effects of intraindividually varying PTSD symptoms on alcohol-related problems in either the logistic (p = .90) or gamma models (p = .86).

American Psychiatric Association. (2000). Diagnostic and statistical manual of mental disorders (4th ed., text rev.). Washington, DC: Author.

Blanco, C., Xu, Y., Brady, K., Pérez-Fuentes, G., Okuda, M., & Wang, S. (2013). Comorbidity of posttraumatic stress disorder with alcohol dependence among US adults: Results from National Epidemiological Survey on Alcohol and Related Conditions. Drug and Alcohol Dependence, 132, 630–638. 10.1016/j.drugalcdep.2013.04.016

Boscarino, J. A., & Figley, C. R. (2009). The impact of repression, hostility, and post-traumatic stress disorder on all-cause mortality: A prospective 16-year follow-up study. Journal of Nervous and Mental Disease, 197, 461–466. 10.1097/NMD.0b013e3181a61f3e

Browne, K. C., Wray, T. B., Stappenbeck, C. A., Krenek, M., & Simpson, T. L. (2016). Alcohol consumption, craving, and craving control efforts assessed daily in the context of readiness to change among individuals with alcohol dependence and PTSD. Journal of Substance Abuse Treatment, 61, 34–41. 10.1016/j.jsat.2015.09.005

Chwastiak, L. A., Rosenheck, R. A., Desai, R., & Kazis, L. E. (2010). The association of psychiatric illness and all-cause mortality in the National Department of Veterans Affairs Health Care System. Psychosomatic Medicine, 72, 817–822. 10.1097/PSY.0b013e3181eb33e9

Cloitre, M., Garvert, D. W., Weiss, B., Carlson, E. B., & Bryant, R. A. (2014). Distinguishing PTSD, complex PTSD, and borderline personality disorder: A latent class analysis. European Journal of Psychotraumatology, 15, 5. 10.3402/ejpt.v5.25097

Coffey, S. F., Saladin, M. E., Drobes, D. J., Brady, K. T., Dansky, B. S., & Kilpatrick, D. G. (2002). Trauma and substance cue reactivity in individuals with comorbid posttraumatic stress disorder and cocaine or alcohol dependence. Drug and Alcohol Dependence, 65, 115–127. 10.1016/S0376-8716(01)00157-0

Curran, P. J., & Bauer, D. J. (2011). The disaggregation of within-person and between-person effects in longitudinal models of change. Annual Review of Psychology, 62, 583–619. 10.1146/annurev.psych.093008.100356

Donovan, D. M. (2003). Assessment to aid in the treatment planning process. In J. P.Allen & V. B.Wilson (Eds.), Assessing alcohol problems: A guide for clinicians and researchers (2nd ed., pp. 125–188). Bethesda, MD: National Institute on Alcohol Abuse and Alcoholism.

Drapkin, M. L., Yusko, D., Yasinski, C., Oslin, D., Hembree, E. A., & Foa, E. B. (2011). Baseline functioning among individuals with posttraumatic stress disorder and alcohol dependence. Journal of Substance Abuse Treatment, 41, 186–192. 10.1016/j.jsat.2011.02.012

Evren, C., Sar, V., Dalbudak, E., Cetin, R., Durkaya, M., Evren, B., & Celik, S. (2011). Lifetime PTSD and quality of life among alcohol-dependent men: Impact of childhood emotional abuse and dissociation. Psychiatry Research, 186, 85–90. 10.1016/j.psychres.2010.07.004

First, M. B., Gibbon, M., Spitzer, R. L., & Williams, J. B. W. (1995). Structured clinical interview for DSM–IV (SCID). New York, NY: New York State Psychiatric Institute.

Foa, E. B., Riggs, D. S., Dancu, C. V., & Rothbaum, B. O. (1993). Reliability and validity of a brief instrument for assessing post-traumatic stress disorder. Journal of Traumatic Stress, 6, 459–473. 10.1002/jts.2490060405

Foa, E. B., & Tolin, D. F. (2000). Comparison of the PTSD Symptom Scale-Interview Version and the Clinician-Administered PTSD scale. Journal of Traumatic Stress, 13, 181–191. 10.1023/A:1007781909213

Fuehrlein, B., Ralevski, E., O’Brien, E., Jane, J. S., Arias, A. J., & Petrakis, I. L. (2014). Characteristics and drinking patterns of veterans with alcohol dependence with and without post-traumatic stress disorder. Addictive Behaviors, 39, 374–378. 10.1016/j.addbeh.2013.08.026

Gaher, R. M., Simons, J. S., Hahn, A. M., Hofman, N. L., Hansen, J., & Buchkoski, J. (2014). An experience sampling study of PTSD and alcohol-related problems. Psychology of Addictive Behaviors, 28, 1013–1025. 10.1037/a0037257

Galatzer-Levy, I. R., Nickerson, A., Litz, B. T., & Marmar, C. R. (2013). Patterns of lifetime PTSD comorbidity: A latent class analysis. Depression and Anxiety, 30, 489–496. 10.1002/da.22048

Goldstein, R. B., Smith, S. M., Chou, S. P., Saha, T. D., Jung, J., Zhang, H., . . .Grant, B. F. (2016). The epidemiology of DSM–5 posttraumatic stress disorder in the United States: Results from the National Epidemiologic Survey on Alcohol and Related Conditions-III. Social Psychiatry and Psychiatric Epidemiology, 51, 1137–1148. 10.1007/s00127-016-1208-5

Grant, B. F., Goldstein, R. B., Saha, T. D., Chou, S. P., Jung, J., Zhang, H., . . .Hasin, D. S. (2015). Epidemiology of DSM–5 alcohol use disorder: Results from the National Epidemiologic Survey on Alcohol and Related Conditions III. Journal of the American Medical Association Psychiatry, 72, 757–766. 10.1001/jamapsychiatry.2015.0584

Hamilton, M. (1960). A rating scale for depression. Journal of Neurology, Neurosurgery & Psychiatry, 23, 56–62. 10.1136/jnnp.23.1.56

Hoffman, L., & Stawski, R. S. (2009). Persons as contexts: Evaluating between-person and within-person effects in longitudinal analysis. Research in Human Development, 6, 97–120. 10.1080/15427600902911189

Jacobsen, L. K., Southwick, S. M., & Kosten, T. R. (2001). Substance use disorders in patients with posttraumatic stress disorder: A review of the literature. The American Journal of Psychiatry, 158, 1184–1190. 10.1176/appi.ajp.158.8.1184

Kaysen, D., Pantalone, D. W., Chawla, N., Lindgren, K. P., Clum, G. A., Lee, C., & Resick, P. A. (2008). Posttraumatic stress disorder, alcohol use, and physical health concerns. Journal of Behavioral Medicine, 31, 115–125. 10.1007/s10865-007-9140-5

King, D. W., Leskin, G. A., King, L. A., & Weathers, F. W. (1998). Confirmatory factor analysis of the clinician-administered PTSD Scale: Evidence for the dimensionality of posttraumatic stress disorder. Psychological Assessment, 10, 90–96. 10.1037/1040-3590.10.2.90

Krause, E. D., Kaltman, S., Goodman, L. A., & Dutton, M. A. (2007). Longitudinal factor structure of posttraumatic stress symptoms related to intimate partner violence. Psychological Assessment, 19, 165–175. 10.1037/1040-3590.19.2.165

Krenek, M., Lyons, R., & Simpson, T. L. (2016). Degree of correspondence between daily monitoring and retrospective recall of alcohol use among men and women with comorbid AUD and PTSD. The American Journal on Addictions, 25, 145–151. 10.1111/ajad.12342

Lehavot, K., Stappenbeck, C. A., Luterek, J. A., Kaysen, D., & Simpson, T. L. (2014). Gender differences in relationships among PTSD severity, drinking motives, and alcohol use in a comorbid alcohol dependence and PTSD sample. Psychology of Addictive Behaviors, 28, 42. 10.1037/a0032266

Marshall, B. D. L., Prescott, M. R., Liberzon, I., Tamburrino, M. B., Calabrese, J. R., & Galea, S. (2012). Coincident posttraumatic stress disorder and depression predict alcohol abuse during and after deployment among Army National Guard soldiers. Drug and Alcohol Dependence, 124, 193–199. 10.1016/j.drugalcdep.2011.12.027

Miller, W. R., & Del Boca, F. K. (1994). Measurement of drinking behavior using the Form 90 family of instruments. Journal of Studies on Alcohol Supplemental, 12, 112–118. 10.15288/jsas.1994.s12.112

Oslin, D. W., Cary, M., Slaymaker, V., Colleran, C., & Blow, F. C. (2009). Daily ratings measures of alcohol craving during an inpatient stay define subtypes of alcohol addiction that predict subsequent risk for resumption of drinking. Drug and Alcohol Dependence, 103, 131–136. 10.1016/j.drugalcdep.2009.03.009

Pacella, M. L., Hruska, B., & Delahanty, D. L. (2013). The physical health consequences of PTSD and PTSD symptoms: A meta-analytic review. Journal of Anxiety Disorders, 27, 33–46. 10.1016/j.janxdis.2012.08.004

Palmieri, P. A., Weathers, F. W., Difede, J., & King, D. W. (2007). Confirmatory factor analysis of the PTSD Checklist and the Clinician-Administered PTSD Scale in disaster workers exposed to the World Trade Center Ground Zero. Journal of Abnormal Psychology, 116, 329–341. 10.1037/0021-843X.116.2.329

Paunovic, N., Lundh, L. G., & Öst, L. G. (2002). Attentional and memory bias for emotional information in crime victims with acute posttraumatic stress disorder (PTSD). Journal of Anxiety Disorders, 16, 675–692. 10.1016/S0887-6185(02)00136-6

Petrakis, I. L., Poling, J., Levinson, C., Nich, C., Carroll, K., Ralevski, E., & Rounsaville, B. (2006). Naltrexone and disulfiram in patients with alcohol dependence and comorbid post-traumatic stress disorder. Biological Psychiatry, 60, 777–783. 10.1016/j.biopsych.2006.03.074

Pietrzak, R. H., Goldstein, R. B., Southwick, S. M., & Grant, B. F. (2011). Prevalence and Axis I comorbidity of full and partial posttraumatic stress disorder in the United States: Results from Wave 2 of the National Epidemiologic Survey on Alcohol and Related Conditions. Journal of Anxiety Disorders, 25, 456–465. 10.1016/j.janxdis.2010.11.010

Prochaska, J. O., & Velicer, W. F. (1997). The transtheoretical model of health behavior change. American Journal of Health Promotion, 12, 38–48. 10.4278/0890-1171-12.1.38

Read, J. P., Brown, P. J., & Kahler, C. W. (2004). Substance use and posttraumatic stress disorders: Symptom interplay and effects on outcome. Addictive Behaviors, 29, 1665–1672. 10.1016/j.addbeh.2004.02.061

Read, J. P., Griffin, M. J., Wardell, J. D., & Ouimette, P. (2014). Coping, PTSD symptoms, and alcohol involvement in trauma-exposed college students in the first three years of college. Psychology of Addictive Behaviors, 28, 1052–1064. 10.1037/a0038348

Sawayama, T., Yoneda, J., Tanaka, K., Shirakawa, N., Sawayama, E., Ikeda, T., . . .Miyaoka, H. (2012). The predictive validity of the Drinking-Related Cognitions Scale in alcohol-dependent patients under abstinence-oriented treatment. Substance Abuse Treatment, Prevention, and Policy, 7, 17. 10.1186/1747-597X-7-17

Saxon, A. J., & Simpson, T. L. (2015). Co-occurring substance use disorders and PTSD. In N. C.Bernardy & M. J.Friedman (Eds.), A practical guide to PTSD treatment: Pharmacological and psychotherapeutic approaches (pp. 135–150). Washington, DC: American Psychological Association.

Schaumberg, K., Vinci, C., Raiker, J. S., Mota, N., Jackson, M., Whalen, D., . . .Coffey, S. F. (2015). PTSD-related alcohol expectancies and impulsivity interact to predict alcohol use severity in a substance dependent sample with PTSD. Addictive Behaviors, 41, 41–45. 10.1016/j.addbeh.2014.09.022

Searles, J. S., Helzer, J. E., Rose, G. L., & Badger, G. J. (2002). Concurrent and retrospective reports of alcohol consumption across 30, 90 and 366 days: Interactive voice response compared with the timeline follow back. Journal of Studies on Alcohol, 63, 352–362. 10.15288/jsa.2002.63.352

Searles, J. S., Helzer, J. E., & Walter, D. E. (2000). Comparison of drinking patterns measured by daily reports and timeline follow back. Psychology of Addictive Behaviors, 14, 277–286. 10.1037/0893-164X.14.3.277

Simpson, T. L., Stappenbeck, C. A., Luterek, J. A., Lehavot, K., & Kaysen, D. L. (2014). Drinking motives moderate daily relationships between PTSD symptoms and alcohol use. Journal of Abnormal Psychology, 123, 237–247. 10.1037/a0035193

Stappenbeck, C. A., Luterek, J. A., Kaysen, D., Rosenthal, C. F., Gurrad, B., & Simpson, T. L. (2015). A controlled examination of two coping skills for daily alcohol use and PTSD symptom severity among dually diagnosed individuals. Behaviour Research and Therapy, 66, 8–17. 10.1016/j.brat.2014.12.013

Tripp, J. C., & McDevitt-Murphy, M. E. (2015). Emotion dysregulation facets as mediators of the relationship between PTSD and alcohol misuse. Addictive Behaviors, 47, 55–60. 10.1016/j.addbeh.2015.03.013

Tripp, J. C., Meshesha, L. Z., Teeters, J. B., Pickover, A. M., McDevitt-Murphy, M. E., & Murphy, J. G. (2015). Alcohol craving and demand mediate the relation between posttraumatic stress symptoms and alcohol-related consequences. Experimental and Clinical Psychopharmacology, 23, 324–331. 10.1037/pha0000040

Vythilingam, M., Blair, K. S., McCaffrey, D., Scaramozza, M., Jones, M., Nakic, M., . . .Blair, R. J. (2007). Biased emotional attention in post-traumatic stress disorder: A help as well as a hindrance?Psychological Medicine, 37, 1445–1455. 10.1017/S003329170700092X

Wilson, S. M., Krenek, M., Browne, K., Yard, S. S., & Simpson, T. L. (November, 2015). Daily effects of PTSD severity and alcohol consumption on self-appraised drinking problems in a PTSD sample with comorbid alcohol dependence. Poster session presented at the meeting of the International Society for Traumatic Stress Studies, New Orleans, LA.

Wolitzky-Taylor, K., Bobova, L., Zinbarg, R. E., Mineka, S., & Craske, M. G. (2012). Longitudinal investigation of the impact of anxiety and mood disorders in adolescence on subsequent substance use disorder onset and vice versa. Addictive Behaviors, 37, 982–985. 10.1016/j.addbeh.2012.03.026

Submitted: October 8, 2015 Revised: November 1, 2016 Accepted: November 3, 2016

Depressive Symptoms, Religious Coping, and Cigarette Smoking Among Post-Secondary Vocational Students

Acknowledgement: Karissa D. Horton is now the Cofounder and Principal Consultant at Limetree Research, LLC.

This research was supported by Grant #R03CA130589 from the National Cancer Institute, awarded to the second author, and an American Association for Health Education Will Rogers Institute Fellowship, awarded to the first author.

Although the rate of cigarette smoking by adults in the United States is now half of what it was in 1964 (U.S. Department of Health, 1964), 19.3% of U.S. adults 18 years and older report that they smoke cigarettes (Schiller, Lucas, Ward, & Peregoy, 2012). Tobacco use remains the leading cause of preventable death in the U.S. (Mokdad, Marks, Stroup, & Gerberding, 2004) and individuals dealing with depressive symptoms tend to smoke more than their nondepressed peers (Anda et al., 1990; Kenney & Holahan, 2008). A study of the historical trends of the relationship between depression and cigarette smoking suggests that the depression−cigarette use relationship has only become apparent in recent years as the prevalence rate of cigarette smoking declined and those who continue to smoke are more likely to be depressed (Murphy et al., 2003).

Depressive symptoms are a significant cause of stress with the potential to spin out of control if an individual does not engage in the necessary efforts to cope with their situation (Smith, McCullough, & Poll, 2003). Tomkins’ (1966) negative affect model for smoking suggests that some may smoke to reduce their negative affect enough to deal with and get to the root of their stress, whereas others may use cigarettes to completely sedate themselves without facing their stressors. Perhaps as Murphy et al. (2003) noted, the relief associated with the reduction of depressive symptoms is so beneficial that depressed individuals continue to smoke cigarettes despite the well-known harmful health effects. Yet, not every individual who experiences depressive symptoms smokes, suggesting that other variables may moderate the relationship between depressive symptoms and cigarette use. Therefore, this study examined the unique role of religious coping, over and above nonreligious coping, as a potential moderator of the relationship between depressive symptoms and cigarette use among a racially/ethnically diverse sample of students enrolled in 2-year postsecondary vocational schools.

The more religiously involved an individual is, the less likely she or he is to smoke cigarettes (see Koenig, McCullough, & Larson, 2001). This association has been reported in samples of college students (Oleckno & Blacconiere, 1991), young adults (Whooley, Boyd, Gardin, & Williams, 2002), and older adults (Koenig et al., 1998). Although these studies indicate an inverse relationship between the institutional aspects of religious involvement (e.g., religious service attendance) and cigarette use (see Koenig et al., 2001), relatively less research has examined the unique role of more personal measures of religiousness, such as religious coping, on cigarette use. Nonetheless, drawing on religious coping theory and existing research (Koenig et al., 2001), it is likely that individuals who use positive religious coping may turn to a “divine other” (e.g., God of monotheistic faiths) for support in times of stress and in this way reduce reliance on smoking, whereas those who engage in negative religious coping may not only lack that sense of support but feel a sense of abandonment that increases the possibility of smoking. From this perspective positive religious coping should be directly inversely associated with smoking, whereas negative religious coping should be positively associated with smoking.

Religious coping can be defined as specific means through which some individuals incorporate “a full range of behaviors, emotions, cognitions, and relationships” to deal with a variety of stressful situations (Pargament, Tarakeshwar, Ellison, & Wulff, 2001, p. 498). Positive religious coping includes a search for a spiritual connection, a collaborative relationship with a divine other, and seeking support from a divine other (Pargament, 1999). Conversely, negative religious coping includes a sense that a divine other is punishing oneself for sins as well as a sense of abandonment by a divine other (Pargament, 1999). A variety of health-related outcomes associated with positive and negative religious coping vary in expected ways. For example, positive religious coping is associated with better stress-related adjustment compared with negative religious coping, which is associated with more adjustment problems (Pargament, Smith, Koenig, & Perez, 1998). Personal measures of religiousness may play a particularly salient role for individuals who experience decreased levels of control in their life. Prayer, for example, is a common coping strategy that individuals draw upon when dealing with stress (see Pargament, 1997).

In addition to the direct effects of positive and negative religious coping on cigarette use, we anticipate that both types of religious coping will moderate the association between depressive symptoms and current cigarette use. Feeling a sense of support from a divine other (positive religious coping) may buffer, or lessen the deleterious impact of depressive symptoms on cigarette use. Depressed individuals are more likely than nondepressed individuals to become socially withdrawn (see review by Tse & Bond, 2004), exhibit excessive reassurance seeking (Joiner & Metalsky, 1995), and express neediness for emotional support. Coyne’s (1976) interactional theory of depression suggests that the increased demand on these relationships often leads to diminished levels of support for, and even rejection of, the depressed individual (see Joiner & Metalsky, 1995). Thus, individuals lacking in social support from their “real” relationships may particularly gain from the support, and the sense of control over stressful situations, provided by a collaborative relationship with a perceived divine other. This relationship with a divine other may rival the intensity of relationships with family and friends (Pollner, 1989). As noted by Pollner (1989, p. 93), the divine is often personified as one who will convey a sense of support as well as guidance for individuals, particularly during stressful circumstances. A supportive relationship with a divine other may diminish a depressed individual’s need to smoke cigarettes by filling the void of social support that those with depressive symptoms often experience. In this way, a relationship with a divine other may make up for a lack of relationship or support from family and friends.

Unlike the buffering role of positive religious coping, negative religious coping may exacerbate the influence of depressive symptoms on cigarette smoking. Unsupportive social relationships coupled with feelings of abandonment and/or punishment by a divine other may decrease one’s sense of purpose and meaning in life and worsen existing feelings of negative affect, all of which may increase cigarette smoking among those with depressive symptoms. For example, compared with their peers who reported lower levels of negative religious coping, individuals living with HIV/AIDS who reported higher utilization of negative religious coping strategies subsequently reported significantly higher levels of depressive symptoms and more symptoms associated with HIV/AIDS (Trevino et al., 2010).

Individuals in blue-collar occupations, which rank low on occupational prestige, tend to lack autonomy in their daily work and a sense of control at their job, which increase their risk for experiencing depressive symptoms (Link, Lennon, & Dohrenwend, 1993). These individuals are also more likely than white-collar workers to have higher rates of current smoking (Barbeau, Krieger, & Soobader, 2004). Blue-collar workers may lack the coping resources typically afforded only to those with higher levels of socioeconomic status and thus, may possibly gain a sense of control through a relationship with a divine other (Pollner, 1989). In this way, religious coping could be a coping resource that fills a specific need for blue-collar workers who perceive a lack of control in their life. Vocational students enrolled in postsecondary 2-year vocational/technical school programs, which prepare students for blue-collar occupations (e.g., welding, air-conditioning), also report higher smoking rates than the general adult population (Loukas, Murphy, & Gottlieb, 2007), making vocational students an excellent population for the current study.

A discussion of depressive symptoms, cigarette smoking, and religious coping would not be complete without examination of gender differences. Women are twice as likely as men to suffer from depressive symptoms (Nolen-Hoeksema, 2001) and women report more cigarette smoking in response to negative affect (Brandon & Baker, 1991). It follows, then, that depressive symptoms may predict women’s smoking outcomes to a greater degree than men’s. Women report more religious involvement than men across a variety of indicators (see Beit-Hallahmi & Argyle, 1997). Likewise, women’s stronger propensity to rely upon and cultivate relationships with others, including a divine other (Pollner, 1989), increases the likelihood that the direct and moderating influence of religious coping will be greater for them compared with men. Moreover, societal norms tend to suggest that women are more reliant upon others to deal with the stressors they face, whereas men are expected to be more self-reliant when working through their problems. So, in this sense, women may be more likely than men to seek out, engage in, and benefit from a collaborative relationship with a divine other, whereas men may be more likely to deal with their problems without such a reliance on a divine other (Maynard, Gorsuch, & Bjorck, 2001).

Findings in the literature indicate that religious and nonreligious coping are correlated and that even among those who draw on religious coping to deal with stress, nonreligious coping is also a fundamental part of their coping repertoire (Pargament, 1997). To provide a conservative test of the impact of religious coping on smoking, the overlapping variance associated with nonreligious problem-focused (active coping, use of instrumental social support) and emotion-focused coping (denial, self-blame) was removed. Thus, we tested whether or not religious coping makes a unique contribution to cigarette smoking over and above the contribution of nonreligious coping. Moreover, because age and financial stress are positively correlated with cigarette smoking (Siahpush, Borland, & Scollo, 2003) and because cigarette smoking outcomes vary by race (Kiviniemi, Oromo, & Giovino, 2011), this study also controlled for these variables. In summary, the current study examined whether positive and negative religious coping moderated the relationship between depressive symptoms and the likelihood of current cigarette use, over and above the contributions of age, race, financial stress, and nonreligious problem- and emotion-focused coping. We hypothesized that:
1. Depressive symptoms, negative religious coping, and emotion-focused nonreligious coping would be directly associated with higher levels of past 30-day cigarette use, and positive religious coping and problem-focused nonreligious coping would be directly associated with lower levels of current cigarette use.
2. Positive religious coping would buffer or lessen the impact of depressive symptoms on past 30-day cigarette use, whereas negative religious coping would exacerbate or worsen the negative impact of depressive symptoms on past 30-day cigarette use.

Participants

Study participants were drawn from a larger study comprised of a convenience sample of 1,120 students recruited from 81 required introductory- and advanced-level classes at two 2-year public colleges in Texas. Student enrollment at the participating schools was 2,590 and 9,582, respectively. The schools offer programs in areas including automotive technology, air-conditioning repair, allied health, business management, computer-aided drafting, medical technology, repair and manufacturing technology, and vocational nursing. Of the 1,434 students enrolled in the 81 classes, 1,131 were in class during the survey administration and 1,120 (78% of enrolled students) volunteered to complete the 117-item anonymous self-report Vocational Student Tobacco Use Survey. Students with any missing data were excluded from the analyses (n = 157); thus, the final sample size was 963.

Of the 963 students who had complete data, 46.8% were women, 53.2% were men; 38.5% were White, 23.9% were African American, 32.8% were Hispanic, 4.7% reported another race/ethnicity, and 10 participants did not provide their race/ethnicity. With regard to age (M = 25.30, SD = 8.59) 60.6% of the participants were aged 18 to 24, 25.6% were 25 to 34, 13.8% were older than 35, and 11 participants did not complete this item. Although respondents did not provide their religious affiliation, county-level data assessed in 2000 indicated that in the first county where data were collected religious affiliation was: Evangelical Protestant, 29.9%; Catholic, 25%; Mainline Protestant, 8.4%; Other; 1.1%; Orthodox, 0.2%; and Unclaimed, 35%. In the second county, religious affiliation was: Catholic, 41.2%, Evangelical Protestant, 16%; Mainline Protestant, 6.1%; Other, 1.8%; Orthodox, .09%, Unclaimed, 34.9% (Association of Statisticians of American Religious Bodies, 2000).

Procedure

Upon approval from the Institutional Review Boards (IRBs) at the University conducting the study and the participating college with an IRB, the researchers scheduled administration of the voluntary and anonymous survey during 25 minutes of class time. A total of 40 classes participated in data collection at the first school (30 classes fall 2007, 10 classes spring 2008). At the second school, a total of 41 classes (20 classes fall 2007, nine classes spring 2008, 12 classes summer 2008) participated in data collection. Across both schools, the survey was administered in 81 classes, from fall 2007 to summer 2008.

Measures

Demographic covariates

Students reported basic demographic information including gender, age, and race/ethnicity. One item from Pearlin, Menaghan, Lieberman, and Mullan’s (1981) measure of economic strain was used to assess students’ perceived financial stress. This item measured students’ financial situation “at the end of the month” (1 = some money left over; 2 = just enough to make ends meet; 3 = not enough money to make ends meet). A high score indicates higher levels of perceived financial stress.

Nonreligious coping covariates

Eight items (four subscales) from the 28-item Brief COPE (Carver, 1997) were used to assess how often students engaged in four different types of nonreligious coping strategies. Students were asked to “think about how you try to understand and deal with major problems in your life. To what extent is each involved in the way you cope?” The four types of nonreligious coping were measured with two items each. Two items measured active coping (e.g., “I’ve been concentrating my efforts on doing something about the situation I’m in.”); two items measured instrumental support (e.g., “I’ve been trying to get advice or help from other people about what to do.”); two items measured denial (e.g., “I’ve been refusing to believe that it has happened.”); and two items measured self-blame (e.g., “I’ve been blaming myself for things that happened.”). Each item was scored on a scale from 0 to 3 and a high score indicated greater use of a coping strategy. Based on an exploratory factor analysis and previous research (Carver, Scheier, & Weintraub, 1989), two common types of nonreligious coping—problem-focused (active coping, using instrumental support) and emotion-focused (denial, self-blame) coping—served as control variables in this study. In this sample, Cronbach’s alpha for problem-focused coping was 0.87 and 0.84, for women and men, respectively. Cronbach’s alpha for emotion-focused coping was 0.85 and 0.87, for women and men, respectively.

Religious coping

Five items were selected from the 6-item Brief RCOPE (Pargament, 1999) to assess how often (0 = a great deal to 3 = not at all) students engage in both positive and negative forms of religious coping when dealing with a major life problem. Prior to answering the religious coping items, students were asked to “think about how you try to understand and deal with major problems in your life. To what extent is each of these involved in the way you cope?” Three items measured positive religious coping (e.g., “I work together with God as partners to get through the hard times.”) and two items measured negative religious coping (e.g., “I feel that stressful situations are God’s way of punishing me for my sins or lack of spirituality.”). The self-directing negative religious coping item (“I try to make sense of the situation and decide what to do without relying on God.”) is considered to be conceptually distinct from the other negative religious coping items and thus, was removed. That is, whereas the two remaining negative religious coping items consider the individual’s relationship with God, the self-directing negative religious coping item taps into the individual’s actions apart from a relationship with God and in this way may be more similar to other nonreligious coping strategies. Each item was reverse coded so that higher scores indicate greater use of positive and negative religious coping strategies.

Although past research studies generally report good reliability for positive religious coping (e.g., α = .81), Cronbach’s alpha tends to be lower for negative religious coping (e.g., α = .58; see Pargament et al., 2001). Consistent with Pargament’s study, the Cronbach’s alpha for the 3-item positive religious coping subscale was 0.86 and 0.90, for women and men, respectively, whereas Cronbach’s alpha for the 2-item negative religious coping subscale was 0.59 and 0.65, for women and men, respectively.

Depressive symptoms

The Center for Epidemiologic Studies Depression Scale (CES-D; Radloff, 1977) is a 20-item measure of the frequency and severity of depressive symptoms occurring in the past week. Students indicated how often in the past week they experienced symptoms of the following four subscales: somatic complaints, depressive affect, positive affect, and interpersonal problems. Each item (e.g., “I felt that I could not shake off the blues even with help from my family or friends.”) is scored on a scale ranging from 0 = rarely or none of the time; less than 1 day to 3 = most or all of the time; 5 to 7 days; therefore, a high score indicates elevated levels of depressive symptomatology. Previous studies have validated the four-factor structure of the CES-D in adult samples (Golding & Aneshensel, 1989) and provided evidence of reliability for this well-established measure of depressive symptoms. Cronbach’s alpha for the CES-D measure was 0.85 and 0.83 for women and men, respectively, indicating that the measure has good internal consistency reliability in the current sample.

Quantity−frequency of current cigarette use

Current cigarette use was measured using two items from the National College Health Risk Behavior Survey (Centers for Disease Control & Prevention, 1997). One item measured frequency of current cigarette use (e.g., “During the past 30 days, on how many days did you smoke cigarettes?”). Students identified how often they smoke cigarettes on a scale ranging from 0 = 0 days to 6 = all 30 days. One item measured quantity of current cigarette use (e.g., “During the past 30 days, on the days you smoked, how many cigarettes did you smoke per day?”). Students identified the quantity of current cigarette use on a scale ranging from 0 = I did not smoke cigarettes during the past 30 days to 6 = more than 20 cigarettes per day. Similar to Schleicher, Harris, and Catley (2009), the two current cigarette use items were multiplied together to create a quantity−frequency variable for cigarette use, which provided a count of the quantity−frequency of cigarettes smoked during the 30 days prior to the survey and ranged from 0 to 36.

Missing Data Analysis

Data from 157 students were removed from the current sample because they were missing substantial amounts of data. A series of analyses were conducted to examine if the data from these 157 students differed from the 963 that were retained for study analyses, potentially resulting in an exclusion bias. The nonparametric Mann–Whitney U test indicated that the distribution of scores on the quantity−frequency of cigarette smoking variable did not differ between the two groups (p = .87). Similarly, independent samples t tests indicated that there were no differences between the two groups on positive religious coping [t(1040) = 1.79, p = .08], negative religious coping [t(1038) = 1.17, p = .24], or depressive symptoms [t(1083) = 1.42, p = .16]. Thus, removal of 157 students’ data from the current sample did not result in exclusion bias.

Data Analysis

Negative binomial regression was used to examine the unique contributions of positive and negative religious coping to the quantity−frequency of past 30-day cigarette use after accounting for the contributions of the covariates (age, race, perceived financial stress, problem- and emotion-focused coping). Negative binomial regression is the preferred statistical method when the dependent variable represents a count of events that are non-normally distributed (Hox, 2010). In the present study, 70.2% of participants received a 0 (indicating they were nonsmokers) on the quantity−frequency variable; thus, this count variable was non-normally distributed.

To determine if the association between depressive symptoms and current cigarette use varied (was moderated) by use of positive and negative religious coping, two separate two-way interactions between depressive symptoms and religious coping were tested. In order to determine the nature of a significant interaction, the methods outlined by Aiken and West (1991) were used. Specifically, the relationship between depressive symptoms and quantity−frequency of past 30-day cigarette use was examined at high (1 SD above the mean value) and low (1 SD below the mean value) levels of the positive or negative religious coping variable. In order to prevent problems with multicollinearity, the nonreligious coping (problem- and emotion-focused) covariates and the depressive symptoms and religious coping (positive and negative) predictor variables were each mean-centered (Aiken & West, 1991).

Each of the covariates (age, race, financial stress, problem- and emotion-focused coping) and all three predictor variables (depressive symptoms, positive and negative religious coping) were entered simultaneously in model one. Both of the two-way interaction terms (depressive symptoms x positive religious coping; depressive symptoms x negative religious coping) were tested independently in model two, with all covariates and main effects included.

Preliminary Analyses

Given that the comorbidity of cigarette smoking and depression vary across males and females (Husky, Mazure, Paliwal, & McKee, 2008), Box’s M test was used to determine the homogeneity of variance−covariance matrices of the study variables across gender in this study. Because Box’s M test was significant (Box’s M = 58.67, df = 36, p = .01) indicating that the variance−covariance matrices varied across gender, all subsequent models were run separately for male and female students.

Mean Differences Across Gender

Mean differences across gender in the study variables were examined for descriptive purposes and are presented in Table 1. With the exception of age, race, and emotion-focused nonreligious coping, results indicated a statistically significant difference at the p < .05 level in scores for each study variable across gender. Female students reported significantly more perceived financial stress, depressive symptoms, positive religious coping, and problem-focused coping than did male students. However, male students reported significantly more negative religious coping and quantity−frequency of past 30-day cigarette use than did female students.

Means, Standard Deviations, and Zero-Order Correlations (R) for Male (n = 512) and Female (n = 451) Students’ Study Variables

Zero-Order Correlations

Prior to testing the study hypotheses, zero-order correlations between the independent and dependent variables for males and females were examined. As shown in Table 1, compared with their non-White peers, both White male and White female students were more likely to smoke cigarettes in the past month. Perceived financial stress, depressive symptoms, and problem-focused nonreligious coping were each associated with increased smoking among female students. Lastly, positive religious coping was associated with decreased cigarette smoking in the past month among female students.

Main Effects

The negative binomial regression, for model one, predicting quantity−frequency of past 30-day cigarette use from the covariates, depressive symptoms, and positive and negative religious coping was statistically significant for male, χ²(8) = 49.08, p < .001, and female, χ²(8) = 159.27, p < .001, students (see Table 2, model one). Among the covariates, age and race were both uniquely associated with past 30-day cigarette use for male and female students, whereas financial stress and emotion-focused coping were uniquely associated with past 30-day cigarette use, but only for female students. Among the predictor variables, depressive symptoms and positive religious coping were significantly associated with quantity−frequency of past 30-day cigarette use, but for females only.

Negative Binomial Regression Analysis for Variables Predicting Quantity-Frequency of Past 30-Day Cigarette Use for Male (n = 512) and Female (n = 451) Students

Given that the negative binomial regression model is log-linear, it is possible to convert the regression coefficients into the predicted multiplicative effect of a 1-unit change in the variable of interest (i.e., depressive symptoms) on the count of cigarette use, holding all other variables constant (Coxe, West, & Aiken, 2009). For example, the exponentiation of the regression coefficient for financial stress (e^0.37 = 1.45) indicates the multiplicative difference in quantity−frequency of cigarettes smoked based on a female student’s financial stress. Thus, a female student with a perceived financial stress score of 3 is expected to have a quantity−frequency score for cigarettes smoked that is, on average, 1.45 times greater than a student with a perceived financial stress score of 2. Examination of the exponentiation of the regression coefficients for the other covariates indicate that for both males and females, a student who is 25-years-old is expected to have a quantity−frequency score that is, on average, 1.02 times greater than a 24-year-old student of the same gender and that White male and female students are expected to have quantity−frequency scores that are, on average, 1.92 and 2.61 times greater than their non-White counterparts, respectively. In addition, a female with an emotion-focused coping score of 9 is expected to have a quantity–frequency score for cigarettes smoked that is, on average, 1.05 times greater than a female with an emotion-focused coping score of 8.

Regarding the exponentiation of the regression coefficients for the female depressive symptoms and positive religious coping predictor variables (see Table 2), results indicate that a student with a depressive symptoms score of 20 is expected to have a quantity−frequency score for cigarettes smoked that is, on average, 1.03 times greater than a student with a depressive symptoms score of 19. Furthermore, a 1-unit increase in positive religious coping causes the expected quantity−frequency score for cigarettes smoked by female students to decrease by a factor of 0.93.

Interaction Effects

As shown in Table 2, model two, four two-way interactions were significant: two between depressive symptoms and positive religious coping, one for males, χ²(9) = 57.08, p < .001, and one for females, χ²(9) = 167.80, p < .001, and two between depressive symptoms and negative religious coping, one for males, χ²(9) = 55.01, p < .001, and one for females, χ²(9) = 171.87, p < .001. Probing the significant 2-way interaction between depressive symptoms and negative religious coping for female students revealed that depressive symptoms were significantly associated with cigarette smoking during the past month at high level of religious coping (β = .05, p < .001), but not low levels of religious coping (β = .01, p > .05; see Figure 1). As expected, findings indicated that negative religious coping exacerbated the influence of depressive symptoms on the likelihood of female students’ past 30-day cigarette use. On the other hand, probing male students’ depressive symptoms x negative religious coping interaction revealed that depressive symptoms were not associated with quantity−frequency of past 30-day cigarette smoking at high levels of negative religious coping (β = .003, p > .10), but were marginally associated with past month smoking at low levels of negative religious coping (β = −02, p = .086). Given the nonsignificant and marginal associations, this finding is not discussed further.

Figure 1. Examining the depressive symptoms × Negative Religious Coping interaction for female students.

Probing the significant depressive symptoms x positive religious coping interaction for female students indicated that depressive symptoms were significantly associated with quantity−frequency of past 30-day cigarette use at high levels of positive religious coping (β = .05, p < .001), but not low levels of positive religious coping (β = .003, p > .05). Although unexpected, these results indicate that high levels of positive religious coping exacerbated the influence of depressive symptoms on the likelihood of females’ current cigarette use. However, examination of the point estimates in Figure 2 indicate that high levels of positive religious coping may be protective for female students reporting low levels of depressive symptoms, but the protective effect is no longer present for those females with high levels of depressive symptoms. For males, probing the depressive symptoms x positive religious coping interaction revealed that depressive symptoms were marginally associated with quantity−frequency of past month cigarette use at high (β = .02, p = .08) and low (β = −.02, p = .07) levels of positive religious coping. Given these findings, this interaction is not further discussed.

Figure 2. Examining the depressive symptoms × Positive Religious Coping interaction for female students.

Despite the growing amount of research examining the influence of religious involvement on various health behaviors, our knowledge of the influence of religious coping on the association between depressive symptoms and cigarette use is lacking. This study extends existing research by examining religious coping as a moderator of the relationship between depressive symptoms and the likelihood of past 30-day cigarette use, while controlling for the influence of nonreligious coping and other covariates in a racially/ethnically diverse sample of vocational/technical school students. In partial support of Hypothesis 1, findings indicated that depressive symptoms were associated with an increased likelihood of cigarette use among female students, whereas positive religious coping decreased the likelihood of female students’ cigarette use, even after controlling for demographic variables and nonreligious coping. Corroborating theory on the role of religious coping, results also showed that negative religious coping exacerbated the influence of depressive symptoms on the likelihood of cigarette use for females. Unexpectedly, however, positive religious coping also exacerbated the influence of depressive symptoms on the likelihood of female cigarette use.

Consistent with Tomkins’ (1966) negative affect model for smoking and prior research (Kenney & Holahan, 2008), depressive symptoms were associated with an increased likelihood of current cigarette use, but only among female students. Past studies report that women have higher rates of depression/depressive symptoms than men (see Nolen-Hoeksema, 2001) and are more likely to use tobacco when dealing with feelings of negative affect (Brandon & Baker, 1991). Moreover, this finding supports those of others showing that depression (Husky et al., 2008) is more strongly associated with cigarette smoking in women than men. The positive association between depressive symptoms and smoking lends itself to the paradox of coping and substance (cigarette) use noted by Brandon, Herzog, Irvin, and Gwaltney, 2004: cigarette smoking functions both as a way to deal with one’s stress and a deleterious health outcome associated with other unproductive efforts to deal with stress.

Additional findings indicated that in comparison with their counterparts, female vocational students who used positive religious coping were less likely to report current cigarette use. Just as Tomkins’ (1966) negative affect model for smoking suggests that some smoke to relieve their stress, findings from this study suggest that female vocational students may use positive religious coping to deal with their stress, but without the health risks of smoking. It is likely that individuals who see themselves as working together with a divine other to deal with their problems may be able to better adapt to their stressful situations (see review by Koenig et al., 2001). These findings add to the growing literature showing the beneficial influence of various aspects of religion on females’ health behaviors (Koenig et al., 2001).

Contrary to the findings for positive religious coping, negative religious coping was not directly associated with past 30-day smoking. However, results confirmed our expectations that negative religious coping would interact with depressive symptoms and exacerbate its impact on the likelihood of current cigarette use. Specifically, female students who reported high levels of depressive symptoms and also engaged in high levels of negative religious coping had an increased likelihood of current cigarette use. Individuals who experience depressive symptoms often exhibit negative emotions and behaviors that turn others away from them, eroding their network of support (Coyne, 1976). Therefore, female students who have strained social relationships associated with depressive symptoms may also feel that God has abandoned them and that their stress is a result of God punishing them. For this reason, female students may be at increased risk for smoking cigarettes.

Unexpectedly, high levels of positive religious coping exacerbated the detrimental impact of depressive symptoms on the likelihood of female students’ current cigarette smoking. In particular, although women with low levels of depressive symptoms may have benefited from the use of positive religious coping, women with high levels of depressive symptoms did not. These findings are partially inconsistent with the direct salutary contribution of positive religious coping to female students’ cigarette use and with our expectation that positive religious coping would buffer the impact of depressive symptoms on current smoking. However, results are consistent with previous findings that positive religious coping exacerbated the influence of perceived racial/ethnic discrimination on the likelihood of African American students’ current cigarette use (Horton & Loukas, 2013). According to Pargament (1997), overreliance on a divine other to assist with or take control of a situation, particularly when an individual needs to take action to deal with their stress, can have harmful consequences. It is possible then that female students with high levels of depressive symptoms may cede personal control over their situation and rely too much on a divine other to get them through difficult situations (Pargament, 1999). Overreliance on positive religious coping to deal with the stress of depressive symptoms may also indicate female students’ inability to assess the best way to deal with their situation. This explanation supports Pargament’s warning of the potential breakdown in the coping process that can occur when the use of religious and nonreligious coping resources are not coordinated appropriately to address a stressful situation (Pargament, 1997). Another explanation for this unexpected finding may be that individuals tend to use all available coping resources, whether positive or negative, to deal with elevated levels of distress associated with depressive symptoms (Coyne, Aldwin, & Lazarus, 1981).

Surprisingly, neither depressive symptoms nor religious coping contributed to the likelihood of males’ smoking. The lack of findings for male students may be due in part to males’ lower mean scores for depressive symptoms and for positive religious coping. The lack of findings for males may also be attributed to the fact that the strength of association between affective disorders and cigarette smoking is weaker for males than females (Brandon & Baker, 1991). Regarding religious involvement, other variables may be better predictors of cigarette smoking for males and should be pursued in future studies. Given that men tend to be less collaborative and less religiously involved than women (Pew Research Center, 2008), subsequent studies should consider the roles of self-directing religious coping (dealing with one’s stress independent of a divine other) and spiritual discontent (feelings of anger and distancing oneself from a divine other) in male vocational school students’ smoking.

The limitations of the current study must be taken into account when interpreting the findings. The cross-sectional study design limits our ability to examine the temporal relationships between the study variables. Future studies should assess both the short- and long-term influences of religious coping on the relationship between depressive symptoms and cigarette smoking using full measures of religious and nonreligious coping. Due to space limitations, a brief form of the religious coping measure and selected subscales of the nonreligious coping measure were used in the present study. Another limitation is the low internal consistency reliability of the 2-item negative religious coping subscale, which may have influenced the study findings. The low internal consistency reliability of negative religious coping may be due to the use of two diverse items measuring struggle with a divine other. Subsequent studies using measures that tap into other more specific aspects of negative religious coping should be conducted to determine if the pattern of findings reported in the present study is replicated. Additionally, the absence of data on the religious affiliation and level of organizational religious involvement of the study participants limits the ability to assess the influence of institutional aspects of religiousness on the study findings. Finally, findings are limited to the examination of religious coping by those affiliated with a monotheistic faith and our measures of religious coping do not include assessment of spiritual coping outside of the influence of organized religion. Future studies should, therefore, examine religious coping for individuals of other faiths and incorporate examination of the role of spirituality in the association between depressive symptoms and smoking.

Notwithstanding these limitations, this study contributes to the literature in several areas, including religious coping, cigarette smoking, and vocational students training to work in blue-collar occupations. This research demonstrated that the direct and moderating influences of religious coping on the likelihood of cigarette smoking occurred in ways that were expected and unexpected. As expected, findings indicated that positive religious coping was associated with a decreased likelihood of female students’ cigarette use and negative religious coping exacerbated the influence of high levels of depressive symptoms on the likelihood of females’ smoking. Although unexpected, this research also suggests that overreliance on positive religious coping increased the likelihood of cigarette smoking for females who reported high levels of depressive symptoms. Simultaneous examination of both the personal (i.e., religious coping, attachment to God) and organizational (i.e., religious affiliation and religious service attendance) measures of religious involvement, depressive symptoms, and cigarette use is needed for a finer-grained analysis of these associations.

There are practical implications to the finding that religious coping is relevant to the female vocational students participating in this study. For females to whom religion is important, religious coping serves as an addition to their repertoire of existing coping resources from which they can draw upon. In particular, such women may also benefit from the social support of those within their religious community and the proscriptive norms of their faith that discourage harmful health behaviors such as smoking (Koenig et al., 2001). The implication of these findings for women who are currently trying to quit smoking, is that in addition to other resources, religious coping may contribute to decreases in smoking.

Aiken, L. S., & West, S. G. (1991). Multiple regression: Testing and interpreting interactions. Newbury Park, CA: Sage.

Anda, R. F., Williamson, D. F., Escobedo, L. G., Mast, E. E., Giovino, G. A., & Remington, P. L. (1990). Depression and the dynamics of smoking. Journal of the American Medical Association, 264, 1541–1545. doi:10.1001/jama.264.12.1541

Association of Statisticians of American Religious Bodies. (2000). Religious congregations and membership in the United States, 2000. Distributed by the Association of Religion Data Archives ( www.theARDA.com). Retrieved from http://www.thearda.com/mapsReports/reports/counties/48029_2000.asp and http://www.thearda.com/mapsReports/reports/counties/48245_2000.asp

Barbeau, E. M., Krieger, N., & Soobader, M. (2004). Working class matters: Socioeconomic disadvantage, race/ethnicity, gender, and smoking in NHIS 2000. American Journal of Public Health, 94, 269–278. doi:10.2105/AJPH.94.2.269

Beit-Hallahmi, B., & Argyle, M. (1997). The psychology of religious behaviour, belief and experience. New York, NY: Routledge.

Brandon, T. H., & Baker, T. B. (1991). The Smoking Consequences Questionnaire: The subjective expected utility of smoking in college students. Psychological Assessment, 3, 484–491.

Brandon, T. H., Herzog, T. A., Irvin, J. E., & Gwaltney, C. J. (2004). Cognitive and social learning models of drug dependence: Implications for the assessment of tobacco dependence in adolescents. Addiction, 99, 51–77.

Carver, C. S. (1997). You want to measure coping but your protocol’s too long: Consider the Brief COPE. International Journal of Behavioral Medicine, 4, 92–100.

Carver, C. S., Scheier, M. F., & Weintraub, J. K. (1989). Assessing coping strategies: A theoretically based approach. Journal of Personality and Social Psychology, 56, 267–283.

Centers for Disease Control and Prevention. (1997). Youth risk behavior surveillance: National college health risk behavior survey–United States, 1995. Morbidity and Mortality Weekly Report, 46(6), 1–56.

Coxe, S., West, S. G., & Aiken, L. S. (2009). The analysis of count data: A gentle introduction to Poisson regression and its alternatives. Journal of Personality Assessment, 91, 121–136.

Coyne, J. C. (1976). Toward an interactional description of depression. Psychiatry, 39, 28–40.

Coyne, J. C., Aldwin, C., & Lazarus, R. S. (1981). Depression and coping in stressful episodes. Journal of Abnormal Psychology, 90, 439–447.

Golding, J. M., & Aneshensel, C. S. (1989). Factor structure of the Center for Epidemiologic Studies Depression Scale among Mexican Americans and non-Hispanic Whites. Psychological Assessment, 1, 163–168.

Horton, K. D., & Loukas, A. (2013). Discrimination, religious coping, and tobacco use among White, African American, and Mexican American vocational school students. Journal of Religion and Health, 52, 169–183.

Hox, J. J. (2010). Multilevel analysis: Techniques and applications (2nd ed.). New York, NY: Routledge.

Husky, M. M., Mazure, C. M., Paliwal, P., & McKee, S. A. (2008). Gender differences in the comorbidity of smoking behavior and major depression. Drug and Alcohol Dependence, 93, 176–179.

Joiner, T. E., & Metalsky, G. I. (1995). A prospective test of an integrative interpersonal theory of depression: A naturalistic study of college roommates. Journal of Personality and Social Psychology, 69, 778–788.

Kenney, B. A., & Holahan, C. J. (2008). Depressive symptoms and cigarette smoking in a college sample. Journal of American College Health, 56, 409–414. doi:10.3200/JACH.56.44.409-414

Kiviniemi, M. T., Orom, H., & Giovino, G. A. (2011). Psychological distress and smoking behavior: The nature of the relation differs by race/ethnicity. Nicotine & Tobacco Research, 13, 113–119.

Koenig, H. G., George, L. K., Cohen, H. J., Hays, J. C., Larson, D. B., & Blazer, D. G. (1998). The relationship between religious activities and cigarette smoking in older adults. The Journals of Gerontology: Series A, 53, M426–434.

Koenig, H. G., McCullough, M. E., & Larson, D. B. (2001). Handbook of religion and health. New York, NY: Oxford University Press.

Link, B. G., Lennon, M. C., & Dohrenwend, B. P. (1993). Socioeconomic status and depression: The role of occupations involving direction, control, and planning. American Journal of Sociology, 98, 1351–1387.

Loukas, A., Murphy, J. L., & Gottlieb, N. H. (2007). Cigarette smoking and cessation among trade or technical school students in Texas. Journal of American College Health, 56, 401–407.

Maynard, E., Gorsuch, R., & Bjorck, J. (2001). Religious coping style, concept of God, and personal religious variables in threat, loss, and challenge situations. Journal for the Scientific Study of Religion, 40, 65–74.

Mokdad, A. H., Marks, J. S., Stroup, D. F., & Gerberding, J. L. (2004). Actual causes of death in the United States, 2000. Journal of the American Medical Association, 291, 1238–1245. doi:10.1001/jama.291.10.1238

Murphy, J. M., Horton, N. J., Monson, R. R., Laird, N. M., Sobol, A. M., & Leighton, A. H. (2003). Cigarette smoking in relation to depression: Historical trends from the Stirling County Study. American Journal of Psychiatry, 160, 1663. doi:10.1176/appi.ajp.160.9.1663

Nolen-Hoeksema, S. (2001). Gender differences in depression. Current Directions in Psychological Science, 10, 173–176. doi:10.1111/1467-8721.00142

Oleckno, W. A., & Blacconiere, M. J. (1991). Relationship of religiosity to wellness and other health-related behaviors and outcomes. Psychological Reports, 68, 819–826.

Pargament, K. I. (1997). The psychology of religion and coping: Theory, research, practice. New York, NY: Springer.

Pargament, K. I. (1999). Religious/spiritual coping. In Fetzer Institute/National Institute on Aging Working Group (Ed.), Multidimensional measurement of religiousness/spirituality for use in health research (pp. 43–56). Kalamazoo, MI: Fetzer Institute.

Pargament, K. I., Smith, B. W., Koenig, H. G., & Perez, L. (1998). Patterns of positive and negative religious coping with major life stressors. Journal for the Scientific Study of Religion, 37, 710–724.

Pargament, K. I., Tarakeshwar, N., Ellison, C. G., & Wulff, K. M. (2001). Religious coping among the religious: The relationships between religious coping and well-being in a national sample of Presbyterian clergy, elders, and members. Journal for the Scientific Study of Religion, 40, 497–513.

Pearlin, L. I., Menaghan, E. G., Lieberman, M. A., & Mullan, J. T. (1981). The stress process. Journal of Health and Social Behavior, 22, 337–356.

Pew Research Center. (2008). Pew forum on religion & public life: U.S. religious landscape survey. Retrieved from http://religions.pewforum.org/pdf/report-religious-landscape-study-full.pdf

Pollner, M. (1989). Divine relations, social relations, and well-being. Journal of Health and Social Behavior, 30, 92–104. doi:10.2307/2136915

Radloff, L. S. (1977). The CES-D Scale: A self-report depression scale for research in the general population. Applied Psychological Measurement, 1, 385–401.

Schiller, J. S., Lucas, J. W., Ward, B. W., & Peregoy, J. A. (2012). Summary health statistics for U.S. adults: National Health Interview Survey, 2010. Vital and Health Statistics 10(252). Hyattsville, MD: National Center for Health Statistics.

Schleicher, H. E., Harris, K. J., & Catley, D. (2009). The role of depression and negative affect regulation expectancies in tobacco smoking among college students. Journal of American College Health, 57, 507–512.

Siahpush, M., Borland, R., & Scollo, M. (2003). Smoking and financial stress. Tobacco Control, 12, 60–66.

Smith, T. B., McCullough, M. E., & Poll, J. (2003). Religiousness and depression: Evidence for a main effect and the moderating influence of stressful life events. Psychological Bulletin, 129, 614–636. doi:10.1037/0033-2909.129.4.614

Tomkins, S. S. (1966). Psychological model for smoking behavior. American Journal of Public Health, 56, 17–20. doi:10.2105/AJPH.56.12_Suppl.17

Trevino, K. M., Pargament, K. I., Cotton, S., Leonard, A. C., Hahn, J., Caprini-Faigin, C. A., & Tsevat, J. (2010). Religious coping and physiological, psychological, social, and spiritual outcomes in patients with HIV/AIDS: Cross-sectional and longitudinal findings. AIDS Behavior, 14, 379–389. doi:10.1007/s10461-007-9332-6

Tse, W. S., & Bond, A. J. (2004). The impact of depression on social skills: A review. The Journal of Nervous and Mental Disease, 192, 260–268.

United States Department of Health. (1964). Smoking and health: Report of the advisory committee to the Surgeon General of the Public Health Service (PHS Report No. 1103). Bethesda, MD: U. S. Public Health Service.

Whooley, M. A., Boyd, A. L., Gardin, J. M., & Williams, D. R. (2002). Religious involvement and cigarette smoking in young adults: The CARDIA study. Archives of Internal Medicine, 162, 1604–1610.

Submitted: December 22, 2011 Revised: July 18, 2012 Accepted: October 16, 2012

Developing a Theory Driven Text Messaging Intervention for Addiction Care With User Driven Content

Acknowledgement: The study was funded through a grant from the National Institute on Drug Abuse (R43 DA029359-01A1; Muench). Frederick Muench was founder of Mobile Health Interventions, which received the SBIR to develop the messaging intervention. This paper describes a treatment development process and not the efficacy of a product or service. Dr. Muench is currently President of Mobile Health Interventions, but the company is in the process of being acquired by a non-profit for which Dr. Muench will remain a consultant. His compensation is fixed and is not dependent on revenue or sales.

The mobile phone is considered by many to be the next frontier in health behavior change because it is a vehicle to deliver personal, intelligent, and adaptive health information anywhere in real time (Patrick, Griswold, Raab, & Intille, 2008; Riley et al., 2011). There is a growing body of research on text messaging for a range of health and mental health problems such as diabetes, asthma, obesity and weight loss, tobacco dependence, bulimia, and even for increasing levels of daily functioning for those with schizophrenia (Cole–Lewis & Kershaw, 2010; Fjeldsoe, Marshall, & Miller, 2009; Heron & Smyth, 2010; Krishna, Boren, & Balas, 2009; Pijnenborg et al., 2010). This is due in part to text messaging or short message service (SMS) being the most accessible and common form of mobile communications. Over 95% of phones are SMS capable, and more text messages are being sent and received every day than mobile phone calls. In addition, although there is a large discrepancy in home Internet access between high and low income households and ethnicities, this is not the case with mobile phone use (Pew Internet & American Life Project, 2010), highlighting that SMS is a particularly useful intervention medium for disenfranchised population groups such as substance users (Fjeldsoe et al., 2009; Heron & Smyth, 2010).

Mobile interventions are particularly well suited for addictions continuing care because they can be utilized with individuals in their natural environment and adapt to their needs in real time using just-in-time therapies. Similar strategies have been used in the empirically supported phone-based continuing care literature (e.g., extended case monitoring, assertive continuing care, recovery management check-ups, early warning signs relapse prevention and concurrent recovery monitoring), which focus on minimal contact over longer periods of time, adapting to the current needs of individuals (McKay, 2009). Despite the promise of SMS interventions for addiction and the encouraging results from tobacco cessation studies (Free et al., 2011), there have been no published studies using SMS for addiction continuing care, nor any studies examining methods to develop effective mobile interventions.

More than other mediums of intervention development, our current behavioral theories may not adequately capture the capabilities of behavior change interventions using mobile phones across disorders. Riley and colleagues (2011) highlight that mobile technology's greatest strength is that it offers the opportunity to use dynamic interactive real-time therapies and that our current behavior change intervention theories, possibly aside from self-regulation theory (Carver & Scheier, 1998), need to be adjusted to account for real-time adaptations at the moment level. Numerous studies have revealed that individuals have very different needs at different times of the change process (Herd & Borland, 2009) and that a range of psychological processes have varying importance throughout the change process (Prochaska & Velicer, 1997; Rothman, Baldwin, & Hertel, 2004). Short message service interventions for smoking cessation have integrated these important just-in-time therapies for participants by providing tailored messages based on participant characteristics and their quit date as well as just-in-time adaptations to care, based on ones current level of functioning (e.g., Brendryen, Kraft, & Schaalma, 2010; Rodgers et al., 2005). Understanding these specific and critical events, as well as in-the-moment decision making, can shape future mobile interventions. With careful understanding, interventions can be individually tailored to each person's needs and specific circumstances.